Connexion
ChatGPT's capabilities continue to expand apace, and the latest feature is a fully capable slideshow creator. Announced a few weeks ago, ChatGPT Slides lets you create PowerPoint-style decks from inside your conversations, and it's now appearing for those on the Pro tier and above (meaning right now, you need to spend at least $100 a month to get it).
If you're on the free ChatGPT plan or the $20-a-month Plus tier, you can still get ChatGPT to generate downloadable PowerPoint files from the content of your conversations. However, they'll be more basic than those made in the full version of ChatGPT Slides, and there are no collaboration or native edit options.
ChatGPT Slides is part of OpenAI's push to give users a complete office suite inside the ChatGPT apps. Pages (the word processor) is already available for Pro tiers and above, while a spreadsheet app is on the way—the idea being that you will soon be able to get all your work done without any help from Google or Microsoft.
While Chatgpt Slides is only available on pricey plans right now, it's likely to come to lower subscription tiers in the future. But for now, how does the tool compare to what's available in the free and Plus tiers, not to mention what you can do with Gemini inside Google Slides? I've taken a detailed look.
When it comes to testing out AI tools, it helps to focus on something you're already knowledgeable about so you can spot any mistakes. To test this, I decided to see what ChatGPT could hoover up from the web about the new iPhone Duo. If you're on the free or Plus plans, you can start a conversation, then ask for a slideshow or PowerPoint about it.
Plus accounts also have access to a specific Presentations add-on, available via the + (plus) menu in the prompt box, though I get the impression this feature is in a state of flux since ChatGPT Slides arrived. In any case, you can begin asking about the iPhone Duo specs and early online reactions, and then say "build me a slideshow on the key features and reviews so far".
What I got back on the $20 Plus account was a fairly competent slideshow on Apple's foldable, with accurate information as far as I could see, and a simple series of layouts. It looked like something you might whip up from a template if you were in a rush and didn't have time to properly align all the bullet points or be very selective about which details you included and which you left out. One neat touch is that ChatGPT was able to drop in an official photo of the iPhone Duo at the start of the slideshow, but I had to specifically ask for one.
You can go back and make specific edits to the slideshow, but you have to describe them in the general prompt box—you can't select individual slide elements and make more precise changes. Still, if all you want to do is occasionally convert your ChatGPT conversations into a rough and ready slideshow, there's no real need to upgrade from the free or Plus accounts for now.
Switching to a Pro account, I gave a similar task to ChatGPT Slides hoping for something more comprehensive. To begin with, Slides gets its own dedicated area in the app, so you can start up a blank presentation from scratch—there are some basic image and text features, but the idea is to use prompts to build your slide deck. It's like a really basic Google Slides, with a prompt box.
Creating a slideshow this way feels rather tedious: Typing out what you want to say and then waiting for ChatGPT to do it isn't as intuitive as just throwing in some graphics and text yourself, as you might do in PowerPoint. It also seems to be a bit buggy, at least during the time I spent testing it, with some requests not properly completing.
Starting a standard conversation with ChatGPT and then turning all the information into a deck is much more useful from my perspective. My iPhone Duo slideshow was longer, laid out better, and more informative than the one I created on the Plus plan—there were even charts of battery life and official photos this time.
However, whether I used the free, Plus, or Pro tools, my presentations didn't come across as particularly well thought through. They don't really flow smoothly, either across individual slides or the slideshow as a whole—it feels very much like an algorithm, rather than a person, has decided what information to include and where on the slides to stick it.
Whether you create presentations in chats or through the ChatGPT Slides space, you're able to click on individual elements and do some basic manual resizing and repositioning. You can also request edits via prompts: "Drop in an image of a cat", "make this pop more", "center all of this information", or whatever it is.
ChatGPT Slides does offer collaborative tools, which I wasn't able to test, and I can see that being useful for teams heavily invested in ChatGPT's ecosystem. I can also imagine ChatGPT Slides coming in handy when you've got a lot of information you need to quickly dump into a slideshow that looks vaguely presentable.
In general though, I think I'd much rather stick to the traditional way of making slideshows for now. These old methods are more precise and more manageable, and give you full control over what's shown and where—even if they do take a bit longer. What's more, modern-day presentation tools come with so many templates and other layout aids that it's not difficult to put together something that's easy on the eyes while including all of the information you think is most critical.
Google is, of course, also all the way in on AI (and then some), and if you pay for Google AI Pro ($19.99 a month) or above, you can have Gemini pop up right inside Google Slides to give you some assistance with your presentation-making. To compare the tools, I gave it the same task as I did ChatGPT, asking for a slide deck on the new iPhone Duo.
Gemini actually asked some questions along the way: about wht visual style to use, the audience to aim the slideshow at, and the details to highlight. Though it's always possible to put this sort of information manually into a prompt, this would be a helpful additional to ChatGPT Slides.
What Gemini and Google Slides produced was impressive in terms of the layout and overall vibe of the presentation: It felt much more like a human being had put together the slide deck compared to what ChatGPT had produced, in terms of the information chosen, the way the slides were laid out, and the color picks.
However, Google's result had its faults too: The presentation got the facts on the maximum brightness of the iPhone Duo screen wrong, and produced an "iPhone Duo" graphic that looked like two iPhones from around 2021 stuck together. ChatGPT Slides at least managed to pull official imagery from the Apple website.
You can't attach individual edit prompts to specific parts of your presentation in Google Slides as you can in ChatGPT Slides—you have to mention them in the general prompt box. That said, Google Slides is obviously a much more mature and established slideshow tool compared to the OpenAI equivalent, in terms of its features and layout options (it has been around for a decade, after all).
Given that extra software structure and the lower cost of entry, Gemini and Google Slides is the better option for AI-assisted slideshow making at the moment—though I wouldn't bet against OpenAI catching up. For now, ChatGPT Slides is something that's good to have available if you're already paying a lot for ChatGPT every month, but not a feature that makes it worth your while to swap your AI allegiance.
Gemini 3.1 Pro launched back in February of this year—quite a while ago, in AI terms. Now Google has jumped to the next full number with Gemini 4 Argon, which it says "excels at deep work in complex, real-world tasks."
You can get the new AI model now inside the Gemini app if you're on the AI Pro ($19.99 a month) or AI Ultra ($99.99 or $199.99 a month) plans. Users on the free and AI Plus ($4.99 a month) plans miss out for now, though you can expect a smaller, cut-down version of Gemini 4 Argon to roll out to these tiers eventually.
Google has a long list of areas where the new AI model excels. There's mention of complex engineering tasks (like checking millions of lines of code for efficiency improvements), predicting molecular structures in chemistry, and delivering 3D visualizations. It also enhances knowledge work across legal, finance, science, and coding.
Google says Gemini 4 Argon is now better at advanced physics modeling, picking through genetic databases, protecting software from vulnerabilities, and resolving problems in physical space (handy for robotics development). The new model "answers questions more directly with less hedging, doesn't repeat itself more than necessary, and skips clichés throughout the responses," according to Google.
With many in the industry worried about an AI-driven apocalypse of some form or another, Google has upped its defenses against misuse and jailbreaking. The company has significantly upgraded its blockers for harmful requests, limits on AI agent overreach, and resistance to malicious prompt injections.
All of these stats are things you would expect to read about in a new AI model announcement, whether it's from Google, Anthropic, or OpenAI: "Our new model is better and safer than ever, and particularly good in these specific benchmarks we've chosen." But is it any more helpful for the average user tapping away in the Gemini app?
I don't have any pressing scientific breakthroughs to research or large-scale software optimizations to run, but I have put Gemini 4 Argon to the test against GPT-6 Sol, which is the latest model available to consumers through the ChatGPT app—though you need to be on at least the Plus plan ($19.99 a month) to get to it.
Make no mistake: Both of these new AI models are quite capable. I first prompted Gemini 4 Argon and GPT-6 Sol with the following: "Using information online, create a web app I can run locally that plots the lighthouses of the world on a map. Clicking on a lighthouse should give me some details about it, show how far its light reaches, and display a simple 2D model showing its shape and height (if known). Put a lighthouse keeper in each model pop-up for scale."
At the highest, slowest levels of "thinking," these AIs took a while to get back to me: Nearly 10 minutes for Gemini and almost 26 minutes for ChatGPT. Both hit the same problem—restricted access to map data online—but both were able to make a rudimentary world map of their own when prompted a second time.
The ChatGPT effort, dubbed "Beacon," wasn't bad, apart from a bit of map clunkiness (it clearly needs to work on its cartography skills). It presented plenty of information, laid out everything well, and gave me nice, neat diagrams on each of the lighthouses featured. The day/night mode worked fine, and it was an impressive bit of coding work, even if it did take 26 minutes.
However, the Gemini effort, called "Pharos," was better in almost every respect. The map was more detailed, lighthouses actually showed their light range (with an impressive animation), and there was more information on each location—perhaps due to Google's expertise in web search. Some of the layout choices were a bit off, but Gemini 4 Argon produced what was clearly the superior app in less than half the time.
Next, I fed PDFs of 10 recent scientific studies into each AI model. The prompt: "Give me a brief summary of each of these studies in plain English, with a single sentence each for the main finding, one study strength, one study weakness, and the main implication." While I hadn't read the studies through entirely myself, I had read news stories on all of them, and I lowered the "thinking" level to low.
This time ChatGPT was faster—25 seconds versus Gemini's 55 seconds—and gave what I thought were the better summaries, at least for my purposes. Gemini did well, and its answers were accurate and to the point, but the summaries ChatGPT gave were easier to understand and shorter. ChatGPT also offered some overall thoughts on the most promising study, which Gemini didn't.
Finally, I checked the imagination and creative juices of the models: "With the success of the Apple TV shows Severance, Slow Horses, Silo, and Widow's Bay, give me five pitches for new show ideas on the platform that combine elements from some or all of these shows. Try and be as original as possible and keep each pitch to around 100 words." I put the "thinking" level back up to "high" for this job.
(Is this how Apple TV executives actually get their ideas nowadays? I hope not.) As far as the answers the AIs gave me, Gemini was faster than ChatGPT (1 minute and 28 seconds against 3 minutes and 48 seconds, respectively), and both served up responses that were a bit more literal in their mashing up than I wanted: failed secret service agents and supernatural terrors abounded, which is I suppose what I asked for.
The ideas were mostly "Widow's Bay set somewhere else" or "Slow Horses set somewhere else," and didn't make a whole lot of sense for the most part, but the best Gemini offering was about a Silo-style underground bunker whose operators go through a Severance-style procedure every 24 hours: outside, the bunker protecting it from zombies by night; inside, the bunker planning social activities for residents by day. It's not perfect, but I might watch the trailer at least.
My favorite of the ChatGPT suggestions was a show about a remote, Severance-style weather station whose employees discover they're not recording the weather—they're creating it. Again, it didn't really logically hold up, but overall I found the suggestions from GPT-6 more nuanced, more likely to be an actual show, and better "thought" through. Maybe Gemini should've given itself more time to "think."
This is only a narrow set of quick tests, but, in my view, it seems Gemini is back to, at the very least, an even playing field with ChatGPT. It performed much better on the lighthouse task, though its summarizing and TV show pitching weren't quite as impressive. Access to the best models requires a subscription on both platforms, but if you're able to, it's worth comparing them against each other on the jobs you most commonly need AI for.
Love it or hate it, AI is changing tech at a rapid pace. Case in point: AI is now being used to reverse-engineer big-name software packages. Once complete, developers upload the cloned versions online as open-source tools that anyone can use for free. It seems no company is immune from these dupes, no matter how big it happens to be. That includes Adobe, and its legacy software like Photoshop, which now has a free-to-download AI clone called "PhotoCraft." I gave this clone a shot, and while it isn't perfect, it's remarkable how similar it is to the real deal.
One of the developers leading this particular charge is Brandon Thomas, a seasoned software engineer who has used Claude Opus 5.5 to build copies of Adobe Photoshop, Illustrator, Premiere, Lightroom, Acrobat Pro, After Effects, and InDesign in the Rust coding language. They all get new names—Photoshop is PhotoCraft, Premiere is FilmCraft—but they're very similar in terms of features and layout to what Adobe puts out and charges a monthly fee for. It's another seismic industry shift driven by AI, and Thomas is apparently building clones of Microsoft Office apps next.
According to the developer, Thomas "got bit by the dark pattern 'cancellation fee' one too many times," which prompted the start of the project. "AI lets us take back tech and set the clock back to the pre-shitty era," says Thomas. "But now you can just create 1:1 functional equivalents, clean room, that are highly performant and cross-platform. Nothing is stopping you." In another comment, he declares, "Software is over."
It's not clear exactly how Thomas completed this project, but he has used the term "vibe coding," so it sounds like some kind of AI-powered observe-and-copy is going on here, even with his programming experience. The developer has also suggested this is a more worthwhile endeavor than supporting open source apps such as LibreOffice and GIMP as, in his view, "There is no reason to save broken software."
As Ars Technica points out, reverse engineering is usually tolerated as long as no code is directly copied, but if the new suite of tools too closely matches the look and feel of the apps they're cloning, that could mean legal trouble. The truth is, no one's sure yet—and it brings up again the legality of the AI models themselves, which have been trained on vast amounts of human-made code to get to this level.
Adobe is no doubt taking a very close look at this, but while it's still available, I decided to give the software a try myself. You can download the apps from the ArtCraft website, though they're labeled as "early alpha" and "in development" for now. According to Thomas, 99% feature parity with the Adobe equivalents should be "measured in months and not years".
The ArtCraft web portal offers a launcher app through which you can manage the entire 'Craft' selection of tools, but I focused on PhotoCraft as Photoshop is the only Adobe program I really make much use of. My first impressions: It definitely looks like a Photoshop rip-off.
PhotoCraft has the same color scheme, the same layout of panels, and the same menus in the same order—with the exception of Plugins, which I suppose will take a while. The keyboard shortcuts are the same, as far as I could tell, and the various preferences screens are very similar too—both in terms of the options you have and how they're laid out.
Most of the Photoshop work I do involves cropping and combining images, tweaking their colors and balance, and getting them in fixed formats for articles. I was able to transform my workflow from the Adobe app pretty much wholesale: The same panels, the same dialogs, and the same tools (from the clone stamp to the magic wand).
Plenty of image editors ape Photoshop, of course, because it's the industry leader—Pixlr, anyone?—but this seems on another level entirely. PhotoCraft is also fast and slick, and as far as I could tell, free from bugs. The only bumps I came across were some clunky snapping and selecting operations, which could definitely be improved upon but didn't actually break anything.
I'm by no means a power user of Photoshop, and people with more experience than me have found plenty of issues under the hood. I suspect that when it comes to playing around with more advanced tools and layer functions, cracks might start to show in the way you would expect from AI code trying to copy what another program does. I'm not sure "watch this program and make something that behaves the same" is the best way forward for software development.
That said, for me, PhotoCraft did everything I needed it to in terms of image manipulations, adjustments, and layer control, and did it well. How the app carries on long-term remains to be seen, as AI-written bug fixes and feature upgrades will presumably be piled on top of existing AI-written code—and that's assuming Adobe doesn't get involved.
Personally, I'm not going to cancel my Photoshop subscription: I can afford it, I use it, and I'm not yet ready to trust my projects to something that's been coded by AI. But this is the era we're in now, where AI can look at something—a painting, a website, a movie, a sprawling Adobe software application—and give you a very convincing imitation in not much time at all.
An abundance of AI imagery is something we all have to live with these days—from school posters to restaurant menus—and while there are still certain tells that indicate something was made with AI, they're not always present. In a lot of cases, it can be almost impossible to distinguish an AI image from a real one.
This is a problem for all kinds of reasons, and, as such, the AI companies are attempting to address it. Images produced by Gemini and ChatGPT come with an invisible watermark called SynthID. That way, AI-generated images don't need visible branding in order to identify their origins. Of course, you need access to a detector to know whether this watermark is present. As it happens, Google just launched a standalone SynthID checker tool on the web, free for anyone to use. I gave "SynthID Detector" a shot, and while it largely works as expected, it isn't perfect.
For a while now, Gemini has let you upload an image and ask if it is AI-generated. With the new online portal, it's even easier to check—though you do still need to sign in with your Google account. Once in, you can check a wide variety of files. This checker supports JPG, JPEG, PNG, BMP, WEBP, AVIF, HEIC, HEIF, TIFF, TIF, and GIF image formats; MP4, MOV, and WEBM for videos, and WAV, MP3, OGG, FLAC, AAC, and M4A for audio. As long as a file was made by a SynthID-supporting AI, it should have a watermark. (The SynthID watermark is a standard supported by Nvidia, OpenAI, Kakao, and Google, though OpenAI has its own detection tool.)
Google doesn't say exactly how SynthID works, presumably to make it more difficult to get around, but does confirm the watermark is "robust against common transformations." It also says the watermark is "woven into the content itself", which means pixels for images and video, and frequency components for audio. The company also admits its detection tool isn't infallible. What it calls "heavy modifications" can degrade the watermark sign, as can multiple "transformations" applied to the same image. The recommendation is to use files that are as high quality as possible, so it seems to me that more data makes detection more reliable (similar to AI text detectors).
Google also says the SynthID watermarks are "imperceptible" to humans, so you don't need to look or listen out for them—they're only revealed by an AI detector tool like this. If you're generating images, video, or audio with AI, the watermarking shouldn't impact the quality of the end result in any way.
I started with some easy tests. The detector correctly spotted that images created in Gemini and ChatGPT, and downloaded directly to disk, were indeed fake. It only took a few seconds to confirm, and it also identified the AI tool that had been used to make them.
It also correctly marked one of my own photos as AI-free, although it did hedge its answer a little, saying that it was "unlikely" that the media was made with AI. All the detector could say with confidence was that there was no SynthID watermark, showing that it's this watermark in particular that it's checking for. It's not looking for any other signs of AI.
I then tried screengrabs of the same AI images, and the detector correctly flagged that they came from Google and Anthropic's AI tools. This shows there's something buried deep in the pixels, rather than in the file metadata itself, revealing the SynthID watermark. You can't just copy and paste to get around it. Applying an oil painting filter in Photoshop to an AI image didn't trip up the detector, nor did adding a slight blur. Adjusting color, contrast, and tone made no difference, and the detector still spotted SynthID when I cropped and combined two different AI images: one from Gemini, one from ChatGPT.
Eventually, I did beat the detector. Once, when I heavily pixelated an AI image—at which point I rather suspect my disinformation campaign would be pretty ineffective anyway—and once when I created a composite image, which was 75% real, 25% AI-generated. With less AI to work with, the detector can falter.
On the back of that result, I used ChatGPT to add an extra person into an otherwise genuine group photo, though this time the tool detected SynthID—even though less of the overall image was AI. There might be some work to do in images that are only partly AI, which is understandable as the detector has less data to work with.
There is a quota limit here, which seems to be around 10 files per day—but 10 each for images, videos, and audio files. Once you've hit your limit, you need to wait 24 hours to try again. As far as I could tell, the SynthID Detector is exactly what Google says it is: a useful check to have, but far from a guarantee one way or the other.
Plex has been around for almost 20 years at this point, and has added plenty of features and functions in that time. We've previously covered some of the ways it can be useful for creating your own personal streaming service.
Even if you're a seasoned Plex user, you should find something of use here, as there's everything from tweaking the metadata that the apps use to create your library, to improving the quality of your video streams—with a particular focus on films and TV shows (though Plex is capable of handling music and photos, too).
Your filenames and folder structure should give Plex enough information to correctly label TV shows and movies, but it doesn't always get it right every time. If you spot a mistake then there are a few ways to change it.
From the main page for a show or film, click the three dots and choose Refresh Metadata. This gets Plex to check again with its online sources, so if the data was missing or incorrect the first time around, this may fix it.
Your second option is to head to the listing for a specific movie or show, click the three dots, and then pick Fix Match. This lets you pick from a selection of online search results, and is useful for titles that are very similar to each other or very common.
The final way of fixing metadata is to go to an individual movie or episode, and click the Edit button (the icon that looks like a pencil). This lets you enter manual metadata edits that won't be overwritten by any online databases that Plex references.
As your media library grows and evolves, clutter can start to build up—particularly when it comes to items you've removed, that tend to leave metadata, preview thumbnails, and other bits of digital detritus behind. These leftovers are known as bundles.
You can clear out bundles and keep your library trim in a couple of ways: In the web app, click the three dots next to your media server on the left-hand navigation pane, then choose Manage Server > Clean Bundles.
The same option is in the troubleshooting part of settings on the web. Click the wrench icon (top right) to open Settings, then choose Troubleshooting > Clean Bundles. You can also follow the Scheduled Tasks link on the same page to have Plex do this chore automatically every week for you.
There may be times when you have multiple editions of the same movie in your library—director's cuts, perhaps, or extended editions—and if you have the Plex Pass plan then you can label these as separate entities inside your library
You can either change the filename on disk to have the edition in curly brackets after the title—so "Blade Runner (1982) {edition-Final Cut}.mp4" for example—or click the Edit button (the pencil icon) on a movie listing page and enter the details in the Edition field.
Your Plex content isn't automatically going to come with subtitles attached, and that's where the FileBot tool can help. Get the program installed on your system, sign up for a free account at OpenSubtitles, and you're ready to start looking for your subtitle titles. This is done via the Subtitles panel inside FileBot.
Use the green user icon to log into OpenSubtitles through Filebot, then search for your films and shows via the drop-down at the top and the Find button. When you find the files you need, double-click them to move them down to the lower window—in there, you can right-click, choose Export, and save them in the relevant Plex library folder.
There are other options, like a basic built-in subtitle search that's available if you have the Plex Pass, and the standalone Bazarr app, which gives you more options than either the integrated tool or FileBot, but takes longer to set up.
You don't have to keep your Plex library to yourself, but there are some tricks to sharing it with friends and family. From the web app, click the wrench icon (top right) to open the Settings pane, then choose Manage Library Access on the left.
Click Grant Library Access to let someone else into your media library: You'll need to supply a Plex username or an email address (which will be used to prompt them to sign up for a Plex account). You're then able to select the servers and libraries they can access. If you have a Plex Pass, you can set more limits in terms of age ratings and whether or not your friends and family can download content.
The wrinkles come with remote access. If you as the media server owner have a Plex Pass, then your friends and family can stream content remotely, as well as on the local network; if not, they each need a Remote Watch Pass ($2 a month or $20 a year) if they're not watching on the same local network as you and the server.
Tautulli is one of the best Plex add-ons around, giving you a host of useful stats about your media server—including records of who watched what. Once you've downloaded the installer for Windows or macOS, you'll be taken step-by-step through the process of connecting it to your Plex server.
If you've shared your Plex library with the family, then you can, for example, see what the kids are watching and when, and even the devices they've used and how much bandwidth they've used up along the way.
The plug-in is interesting enough even if you haven't shared your library: It's possible to break down viewing to day-by-day or hour-by-hour graphs, see the movies and shows you're watching most often, check out detailed information on your media files, and keep track of when new films and shows are added.
Collections are a useful way of applying some order to the chaos of your Plex media library, and the feature is perhaps not as well known as it should be. You can collect whatever you like, whether it's feel-good TV episodes or spy films.
To get started with collections, click the Edit button on a listing page (the pencil), then choose Tags, and enter something into the Collections box. If the collection already exists, it'll show up as you start typing.
You can also set up smart collections, where media files are added automatically. Set a search filter via the Library tab on your Films or Shows screen, and if you click the Add to button (the plus and horizontal lines icon on the right), you get a Create Smart Collection option. Any files that are added in the future that match your search filter get automatically added to the collection too.
Most of the time Plex will handle video quality automatically, and do a good job of it, but there might be occasions when you're not getting the best quality possible (particularly if you're watching on less powerful devices).
To make sure Plex is always serving up video as best it can, in the web app click the Settings icon (the wrench, top right), then Transcoder. If you know you've got the computer to handle it, you can set the top Transcoder quality drop-down to Make my CPU hurt, and enable the Use hardware acceleration when available.
In the apps for individual devices, you'll also find video quality options in the settings. Put video quality up to Maximum for best results—you may need to experiment to find the limits of what each device can handle.
You might not have given much thought to the Continue Watching pane that shows up on your Films and Shows pages, but you can customize this on the web to suit your own preferences—and maybe remove some content you don't want to keep around.
Click the wrench icon (top right) to open Settings, then select Library on the left. You're able to set a threshold (a number of weeks) beyond which titles disappear from Continue Watching, as well as limiting the number of titles included in title. Turn off the Include season premieres in Continue Watching checkbox to hide premieres for new seasons when you've finished old ones—otherwise they'll show up indefinitely.
The settings you configure here should propagate to the Continue Watching rows in your other apps; try closing and restarting them if not.
If you miss the traditional channel surfing experience when using Plex, there's a solution in the form of QuasiTV: This app works on Android TV, Google TV, and Fire TV devices, and wraps your Plex library and streaming services in a more conventional interface.
Sign into Plex, point it towards your libraries, and it'll create a custom channel guide filled with your shows and movies. These might get assigned to genres, or to the TV stations they originally aired on, and you can browse around until you find something you like—just like the old days.
It's not the most efficient way of searching through your Plex media library, but it's fun to use, and when I tested it out I rediscovered a few movies and TV episodes that I hadn't watched in a while.
We may earn a commission from links on this page.
Samsung is rolling out a major One UI 9 update, based on Android 17. If you've got one of the Galaxy S26 handsets launched at the start of the year, you should see it pop up on your device soon (it came pre-installed on the Galaxy Z Fold 8 and Galaxy Z Flip 8 devices unveiled in July).
After the Galaxy S26, next in line for the update are the Galaxy S25, Galaxy Z Fold 7, and Galaxy Z Flip 7 phones—though Samsung hasn't said exactly when the software will arrive for those devices. After that, phones from as far back as 2023 should eventually get One UI 9, including the Galaxy S23, Galaxy Z Fold 5, and Galaxy Z Flip 5 handsets.
If you do have an eligible phone, you'll want to dive in and start making the most of the update as quickly as possible. These 10 hacks should get you up and running. From making your device more secure to creating your own custom widgets, these are my favorite tricks and tips for the updated OS.
AI Network Guard is a new feature in One UI 9 that cuts off the internet when you open an app you've put on a specific restricted list. It's actually designed to stop kids from stumbling across unsuitable content, but it works great for limiting doomscrolling for grown-ups as well.
The feature isn't easy to find, though—it's on the same hidden menu that features Samsung's wifi signal scanner (why Samsung is keeping it hidden is anyone's guess). From Settings, tap Connections and Wi-Fi, then tap the three dots in the top right corner and select Intelligent Wi-Fi. On the next screen, tap the Intelligent Wi-Fi button seven times to reveal a new Connectivity Labs option.
Go into Connectivity Labs, and you will at last discover AI Network Guard. Follow the instructions to set up a PIN, and you can then pick the apps you want to block internet access for. Use the Downtime feature (via the three dots, top right) to set certain periods in the day when those apps can be used as normal.
One UI has always impressed in terms of the number of customization options it gives users, and there's a new tweak available in the latest update. If you swipe down from the top right corner of the screen to get to the Quick Panel, then tap the pencil edit button, you've got more freedom to adjust its layout. You've always been able to swap widgets in and out, but now you can position them in any way you like (just tap and drag), and change their sizes (via the handles around the edges). The sound button is also now decoupled from the volume slider, so you can manage them separately. Tap Add a control to introduce new elements, and tap Done when you're finished.
If you want one person in particular to be the star of a video, My FanCam is here for you in One UI 9. It lets you load up a video you've already recorded and create a new clip with a specific individual front and center. The obvious use case for this is your kid in the school play or at sports day, but you can use it for whatever you want. You do need to give My FanCam plenty of material to work with in terms of footage and framing—it can't magically produce new angles or new clips out of nothing.
To get started, open up a video from the Gallery app, then tap the Galaxy AI button (the stars) and My FanCam to get started. After a few moments, you'll be able to choose the person you want to follow and select crop and aspect ratio settings.
Samsung Now Brief is aiming to give you timely alerts throughout the day about information that's relevant to you—such as weather updates or appointment reminders—and in One UI 9 you can make it more useful with the option to create your own cards. From Settings, tap Galaxy AI > Now Brief, then choose Custom cards. You can have three active at any one time, centered around weather details, health statistics, or particular categories of video pulled from the web.
Google has finally gotten around to building something to match the anti-car sickness Vehicle Motion Cues feature that Apple put into iOS last year, and it's called Motion Assist. It was rolled out as part of the Android 17 update, and Samsung has carried it over to its own One UI 9 upgrade.
To find it, open up Settings, then choose Google services > All services > Motion assist. Beyond enabling and disabling the feature, you can customize the color, shape, and opacity of the dots that flow across the screen while you're moving, and which will (hopefully) keep any nausea at bay.
Give your wallpaper game a bit of a lift by setting a Motion Photo as your phone's backdrop, rather than a static image. These images are accompanied by short videos when they're captured, giving you a second or so on either side of the still frame—tap the little icon that's an 'A' inside a series of rectangles to turn it on in the Camera app.
One UI 9 lets you set these Motion Photos as wallpaper to add a bit more movement and interest. You still set the wallpaper in the same way: Wallpaper and style, then Change wallpapers > Gallery from Settings. You should see a Motion photo option appear on compatible images.
The Finder is the search tool built into One UI, and it's traditionally been available at the top or bottom of the app drawer that appears when you swipe up on a home screen: You just type in the app or file you want to find on your phone. One of the nice tweaks introduced with One UI 9 is the option to have the Finder on the home screen for more convenient access. Open Settings, then choose Home screen > Show search on Home screen to enable it.
AI continues to become more integrated into our phones, and Galaxy handsets are no exception: You may well have several third-party AI apps on board your device, as well as the built-in Galaxy AI features. One UI 9 ushers in a new "AI Assistant Activity" dashboard that lets you keep tabs on what AI apps and their agents are doing on your phone, from booking concert tickets to analyzing photos. You can see everything that's been happening in one place. You can find the dashboard from Settings by tapping Security and privacy > AI assistant activity. Any actions AIs have taken on your behalf can be viewed app by app, with filters for activity over the last seven days or the last 24 hours.
Perhaps not a One UI 9 upgrade that's going to grab many headlines, but a useful one nevertheless: Samsung Notes now has a Tape tool that works exactly like the name suggests: It lets you put a bit of digital "tape" over part of your note. You can use this as a revision aid, as a decoration for scrapbooking or journaling, or to protect sensitive information when you're sharing screenshots of your notes. You'll find it along the bottom toolbar when a note is open (the icon looks like a piece of tape). You can choose different colors and patterns for your tape, change the size, and have the app automatically straighten your taping too. Use the tape icon on the tool pop-up to show or hide all your tape.
This is a tip that you'll hopefully not have to call on too often, but there's a new Warranty and Care section in Settings that brings together a few older features and adds some new ones on top. From here, you can sign up for Samsung Care+, check if your phone is still under warranty, book repairs for your device, or run a diagnostics check on your phone. You can even ask Bixby for some troubleshooting tips and advice.
Google launches (and retires) apps at a rapid rate, so it's not always easy to keep track of everything that's coming and going. One of the more recent launches we've had is Google Pics, which is now available to users subscribed to Google AI Pro ($19.99 a month) or Google AI Ultra ($99.99 or $199.99 a month).
It sits alongside Google Docs, Slides, and Sheets in your Google Drive Workspace, and it's part Canva, part Photoshop: It does a bit of graphic design (like putting together an event poster), and a bit of image editing (like cutting unwanted objects out of your vacation photos), and if you're paying for one of those Google AI plans mentioned above, you might find Google Pics is the only image creation app you need.
The web app incorporates the Nano Banana AI image generation model, and, as well as working as a standalone entity, will also pop up inside the other Google Workspace apps as and when you need it. Here's what you get and how it works.
Google Pics is driven primarily through text prompts, whether it's creating something new from scratch or making edits to something you've made elsewhere. From the opening screen, you can type in an image prompt just as you would with the Gemini app: Prompts for anything from photorealistic scenes to printed materials will work, and the more detail you add, the better.
The Style box lets you give Google Pics a reference to work from in terms of the overall image look and feel, while clicking on the Upload, Drive, or Photos links lets you load in an existing image ready for editing. At the very top is an aspect ratio drop-down, so you can specify the dimensions of new images, though this can be edited later too.
When you make something new, you get four or five alternatives shown as thumbnails on the left, so you can click through them to find the one you want. There's also the option to generate four more pictures from your prompt, if you don't mind the extra usage. As it often is, Google is rather vague when it comes to limits: It says "higher access" will be available to users until February 2027, and that factors like image resolution affect how quickly you reach the limits.
So, you could prompt Google Pics like this: "Create an A5 flyer for a 10th birthday party for Jess, happening at the village town hall at 3 p.m.-6 p.m. on Sunday, September 20. Include lots of party decorations, including balloons, streamers, and cake, and include some small print about food and drink being provided and there being plenty of on-site parking."
From that prompt, Google Pics gave me a selection of suitable vibrant and fun-looking flyers I could choose from—all of them carrying that familiar AI "sheen" you've no doubt seen everywhere from restaurant menus to school event posters in recent times. It followed the brief directions and took seconds to do.
I also tried loading in my own photos and using Google Pics to manipulate them: You can very easily turn your photos into watercolors or comics, giving you what's effectively an unlimited box of Photoshop filters to play around with. The results aren't perfect every time, but they're usually more than good enough.
This is a pretty bare-bones app though. You can undo recent edits, zoom in and out, and change the aspect ratio, but outside of the prompt box, that's just about it. You don't have any of the traditional resizing options or tools to do any drawing or touching up yourself, and there's no support for layers—so you're still going to need Photoshop and Canva around for more advanced tasks.
Where Google Pics really impresses is with the editing tools, which become available after you've created something from scratch or loaded in an existing image. First of all, the app identifies parts of the image you can click on—so "church", "sky", and "mountains," for example. You can click on these specific elements to get a prompt box to make changes.
You can also just drag a box around something on screen to bring up a prompt box: Maybe drag a box around a party balloon and add the prompt "make this blue," or select a random stranger in your family vacation photo and say "remove this person". Any edit you can describe, you can make (or at least attempt).
As always with this kind of trickery, Google Pics doesn't know what's actually behind the people and objects it's removing, but it can make a decent AI-inspired guess based on the rest of the image and background. The results I got were mostly fine, and usually you wouldn't know an edit had been made—though bigger changes to more detailed photos, like removing buses and cars, can be difficult for Google Pics to handle.
You can save up all of these various edits and apply them together, which saves quite a bit of waiting and working forwards and backwards. In the case of my birthday party flyer, I was able to add lines of text, change various colors, add in new elements, and more—and if you select blocks of text, you can edit them directly without a new prompt.
Besides being useful practically, it can also be a lot of fun playing around with different edits and seeing what's possible. I got quite addicted to playing around with different photos, making changes that ranged from the large scale ("turn this into a night scene") to the small scale ("add an extra cloud here").
One of the edits I made was to age a car in a scene by five years, something which I've got no idea how to do in a full image editor, and which would've probably taken me an hour or so even if I did. The end results might not be quite as precise or polished as they would be if done by a human hand, but these kinds of tools expand what's possible with quick and casual image editing.
I won't be ditching Canva and Photoshop just yet, as they're still essential for different kinds of tasks, including projects where pixel-level precision is required and where various elements need to be combined together in layers—but I can also see Google Pics proving very useful, especially with its integration with the rest of Workspace.
Tailscale wasn't on my radar until a friend recently raved about its capabilities to me. Looking around the web, it seems the software has attracted a loyal following; perhaps part of the reason I hadn't come across it before is that it's not all that easy to understand what it does. Tailscale is technically a VPN (Virtual Private Network), but not in the way the term is typically used. Whereas something like NordVPN or ExpressVPN connects you to someone else's network to access the wider web, Tailscale does the opposite, and connects you to your own network from outside.
"Remote access technology" has been around for years, of course, but it often involves a lot of detailed and technical setup steps—you want to be able to access your home network remotely without letting anyone else in, and that usually means some complicated configuration involving router settings and several third-party apps.
There's none of that with Tailscale, and when you actually install the software, you quickly start to see what all the fuss is about. You can be sitting in a coffee shop or hotel room and have virtually the same access to your home network as you would if you were actually in your living room, which has all kinds of benefits, from file sharing to managing your smart home devices.
The obvious use case is remote access to a computer you have at home, without the layers of complexity or using another company's servers. Tailscale also makes it easier to access media servers you've set up with apps like Plex or Jellyfin, and smart home setups running through Home Assistant or another dashboard.
Everything is built on the open-source WireGuard VPN protocol, which creates private and encrypted links to all of the devices you want to connect to your network—your "tailnet." To add a new device, just install Tailscale on it, whether it's your phone on a 5G connection or a desktop computer set up at the office. Tailscale is free for personal use, which covers up to six individual users and an unlimited number of personal devices, though there are some limits: 50 tagged resources (always-on devices that need multi-user access, like a NAS box), and three Access Control List or ACL groups (for setting access permissions per user). Paid plans start at $8 per month.
The ease of setup is one of the main appeals of Tailscale, and it couldn't be much more straightforward. Download the relevant app on whatever device you want to start with, create an account (you can sign in using a Google, Microsoft, GitHub, or Apple account), and you're up and running. I had a Windows PC, a MacBook, an Android phone, and an iPhone connected to my tailnet in a matter of minutes, with just a few taps and clicks required. There's also a web portal you can access to see all of the devices you've connected and what they're up to.
One improvement I noticed right away was the addition of Tailscale to the share sheet on Android and iOS, which meant I could quickly share photos and other files between all my devices in seconds (no AirDrop or Android Quick Share required). I could also instantly access the shared folders on my desktop and laptop computers from my phone.
Something else you can do very easily is set up a remote access app on a phone and use it to connect directly to one of your computers in a different physical location, with no exposure to the wider internet and no advanced configuration required—Tailscale provides you with an IP address that looks like a local network. Any of your home computers can be configured as an exit node as well, which means internet traffic can be rerouted through it. If you're in a hotel, for example, you can use Tailscale as a secure tunnel back home, which means websites see you as visiting from home rather than an unrecognized hotel IP address.
There are some areas where Tailscale can improve (the Windows client isn't quite as straightforward to use as the macOS one), and it's fair to say at least some basic networking knowledge is required to get to grips with Tailscale, but in terms of software that's free and useful, it's hard to beat. I didn't dive into more advanced use cases involving NAS drives or home servers, as they're not something that I'm working with at the moment, but that's the next level of Tailscale capability
We may earn a commission from links on this page.
After years of rumors, Apple finally made a foldable: the iPhone Duo. The initial reaction online has been mostly positive—particularly how iOS adapts to the phone and that cool transition wallpaper effect—but it begs the question: How does Apple's new device compare to the folding phones already on the market?
The iPhone Duo is now competing against the Google Pixel 11 Pro Fold and the Samsung Galaxy Z Fold 8 (which I'll focus on here rather than the Z Fold 8 Ultra). There are key similarities and differences between the three handsets that suggest whether or not the iPhone Duo will be a success. And having used both Google and Samsung foldables in the past, I have some ideas about what the iPhone Duo needs to get right.
Google and Samsung have been doing this longer than Apple has, and their phones are thinner and lighter than the iPhone Duo. That might make a difference once Apple's handset gets out into the world, with preorders set to start on Friday, Oct. 16. You can also see here that Apple and Samsung have gone for the "passport" style approach: Displays that are shorter and wider, which means a more tablet-style main display. The Google Pixel 11 Pro Fold has the aspect ratio of a standard smartphone when closed, with a main display that's more square.
You can see how the passport-style foldable compares to the more traditional foldable form factor in the display dimensions too: There's a lot more vertical space on the Pixel 11 Pro Fold when all of these foldables are closed shut, which is quite a consideration. If you're like me, you'll end up using your foldable closed more often than unfolded, in which case the outer display becomes very important. (But more on that later.) It's worth knowing that both Apple and Google rely on Samsung to supply their phone displays, as per industry sources.
If you'd like a foldable phone, it's going to cost you, whichever brand you favor. The iPhone Duo starts a little higher than the competition and goes up to 2TB of storage—something Google and Samsung don't do. It's worth mentioning RAM here too: The iPhone Duo has 12GB, the Google phone comes with 16GB, and Samsung offers either 12GB or 16GB in the Galaxy Z Fold 8, depending on the amount of storage you buy. But it's not an even comparison: The Duo runs iOS, which has different RAM requirements than Android devices.
Cameras abound on foldable phones—you need a selfie camera for the inner display and the outer display—and it's notable that Apple isn't telling anyone the specs of its inner selfie cameras. Google's foldable is the only one with a telephoto camera. In the modern era, software processing is just as important as hardware specs, so how the handsets compare on this front is to some extent subjective. It'll be interesting to see how photos from the iPhone Duo stand up against those on the iPhone 18 Pro, which has a more advanced camera system.
The iPhone Duo doesn't have face unlock as the Google and Samsung phones do, and the Galaxy Z Fold 8 doesn't have the IP68 waterproofing of the iPhone Duo and Pixel 11 Pro Fold (though this doesn't last). Battery life is a key consideration when picking a smartphone, but the capacity numbers here can only tell us so much. The Pixel 11 Pro Fold has a 4,806 mAh battery, the Galaxy Z Fold 8 has a 4,800 mAh battery, and Apple isn't revealing the battery capacity on the iPhone Duo—though it does say you can get up to 44 hours of video playback between charges on the outer display. Software optimization and other factors play a role too, so some real-world testing will be needed for a proper comparison.
Speaking of software, there's the iOS and Android divide, of course. From what we've seen of iOS on the iPhone Duo, it looks like an elegant and thoughtful adaptation, with shifting buttons, split keyboards, a dual pane mode, and some other nice touches—no doubt helped by Apple's iPad experience. To my eye, it seems like a more fluid experience than what you get on an Android foldable, though that's not to say you get a bad experience on Google and Samsung phones.
Spec sheets can only get you so far, and while the iPhone Duo has made a bold impression based on what we've seen from Apple so far, no one has properly used one yet. The true test for whether the foldable iPhone is a hit or not is going to depend on how well it works through all the moments of day-to-day life.
I've spent several years using foldable phones from Samsung and Google, and aside from the rather prohibitive cost, the main issue that makes me still prefer standard phone form factors is that, for most of the time, there's no reason to open up the full display—which means you're left with a device that's chunky and heavy, with no real upside. Checking messages on the go, making contactless payments, taking photos: the majority of phone tasks don't need or suit a tablet-sized screen. That's what all of these foldables are fighting against. Would you use an iPad to take a photo at a gig, call your parents, or find your way around in an unfamiliar city?
Apple seems to have the software experience right, and the iPhone Duo looks as well crafted as you would expect a foldable iPhone to be—but even if you can foot the bill for one of these, you might find that the cost of using a thicker, heavier, smaller phone isn't worth it, just so you can occasionally have a bigger screen available. The iPhone Duo actually stacks up pretty well against the foldables from Samsung and Google, but its real competition is going to come from other iPhones.
We may earn a commission from links on this page.
There was something different about today's "Surprise and Shine" launch event from Apple: We didn't get the unveiling of the iPhone 18. For the first time since Apple started releasing multiple iPhones per year, which was 2013 with the arrival of the iPhone 5c and the iPhone 5s (simpler times), we have several pro-level handsets without the standard version to accompany them. So what's behind the shift? And is the iPhone 18 canceled? Apple, predictably, isn't saying, but we've got enough clues to make an informed guess.
If you've been keeping up with Apple rumors since the launch of the iPhone 17, you'll know that this has long been predicted—notably by the most respected purveyor of Apple leaks in the business, reporter Mark Gurman. Gurman was predicting in Bloomberg as far back as November 2025 that we would only get the iPhone 18 Pro, iPhone 18 Pro Max, and the folding iPhone Duo in 2026.
The reason? According to Gurman, it's to spread out revenue more evenly across the year, rather than having most of the iPhones unveiled every September. It also takes some of the pressure off Apple's marketing teams and reduces the likelihood of mishaps like that Siri AI upgrade—which was announced way before it was ready. Gurman also went on record as saying that this is likely to be the norm for "years to come," so get used to it.
However, Gurman wasn't the first to break the news. In May of 2025, Wayne Ma at The Information reported that Apple would hold the standard iPhone 18 model over until 2027, based on conversations with anonymous supply chain sources. As per that report, one of the motivations behind the shift was to make more space for the folding iPhone, which makes sense considering it's Apple's first attempt at the form factor.
Just a day later, well-known and usually well-informed Apple analyst Ming-Chi Kuo backed up the story, saying that we'd see the foldable iPhone, the iPhone 18 Pro, the iPhone 18 Pro Max, and the iPhone Air 2 (then called the iPhone 18 Slim) in September 2026. Apart from that last handset, he was right.
Kuo added an extra reason for the shift as well: Competition from phone manufacturers based in Asia. A lot of these manufacturers release their flagship handsets during the first half of the year, including Samsung, and as far as Apple is concerned, that can leave it playing catch-up by the time September rolls around.
While the iPhone 18 was a no-show at the "Surprise and Shine" event, there's no indication that it's been canceled. Apple still wants to cater to those who'd rather spend a little less on their iPhones (the iPhone 18 Pro starts at $1,119), and the most educated guesses we've seen suggest we'll get the base version of the phone sometime in March.
It's unlikely to be on its own either. We're also expecting to see the iPhone 18e and the iPhone Air 2 appear at the same time, following on from the iPhone 17e (launched in March 2026) and the original iPhone Air (launched in September 2025, alongside the iPhone 17 series).
And that "e" phone series set the foundation for the calendar shift we've now seen with the standard iPhone 18 model. The first of these more affordable handsets, the iPhone SE, showed up in March 2016—it was then followed by the iPhone SE 2 (April 2020), the iPhone SE 3 (March 2022), and the iPhone 16e (February 2025). That shows Apple has been eyeing up the February-March-April slot for non-pro-level iPhones for a decade at this point.
The rumors so far suggest the iPhone 18 will use an A20 chipset, while we're likely to get a RAM bump as well. We now know that the iPhone 18 Pro and iPhone 18 Pro Max have a smaller Dynamic Island, with the ability to show up to three Live Activities, and that's something that should carry across to the standard model (along with iOS 27).
Apple didn't mention the iPhone 18 at all today—not even in a post-credits sequence—but it would be a major surprise if the handset didn't make a belated appearance sometime in the first quarter of next year. If this is the iPhone that's for you, you're just going to have to wait a little while longer.
In among the maelstrom of everything else that's happening with AI right now, OpenAI has pushed out ChatGPT Images 2.5, which it describes as faster, sharper, and smarter, with "better tools for creating whatever you can dream of." Those new tools include templates, more precise editing features, and a way of converting your on-screen doodles into fully realized images.
This Tweet is currently unavailable. It might be loading or has been removed.
Does all of this mean ChatGPT Images 2.5 will save us from the scourge of bad AI food images on menus and posters? Can it replace a traditional photo editor? And how does it compare to the impressive Nano Banana 2 update that Google pushed out in February?
To answer all these questions and more, I fired up the new image generator in ChatGPT to see what it was capable of. Everyone is getting it—it's rolling out for all users across the web, the desktop apps, and the mobile apps now—though as usual there are varying usage limits (which ChatGPT only vaguely defines) on each plan.
The sameness and uncanny look of AI-generated food on restaurant signs and menus is becoming a real issue, so this seemed like a good place to start. It's fair to say that promotional images of food have always taken some artistic license, but AI is making the problem worse. Can ChatGPT Images 2.5 help?
Here's what I tried: "Create a 1:1 aspect ratio image with four squares inside it for a fast food restaurant display. The images should show (from top to bottom, left to right) a cheeseburger, six nuggets, a box of fries, and a Coke. Make the pictures look as photorealistic as possible with natural lighting, but add variations and imperfections to give the impression of real food."
ChatGPT is on the left and Gemini on the right. Both do a decent job of producing menu items that look realistic and actually edible, albeit with the usual AI tendency towards the generic. Personally, I'd say the ChatGPT images look slightly more like real food, though Gemini has done better on the backgrounds and wider scene (the Burger King logo would cause trouble, though).
One of the key new features in ChatGPT Images 2.5 is a new Sketch tool, which lets you turn your basic scribblings into full pictures. Click or tap the + (plus) button in the prompt box to find it—if you're on mobile you'll also have to choose Plugins from the sub-menu that pops up.
The tool lets you play around with colors, shapes, and text, and when you're done you can add a text prompt to it—something like "turn this sketch into the sort of illustration you might see in a high-quality children's book, complete with stars and distant planets" (which is the prompt I used here).
Gemini doesn't have a Sketch tool per se, though it's easy enough to add an image to a prompt for inspiration. Google's AI went far more traditional with its illustration and decided to drop in an alien astronaut, whereas ChatGPT used a touch more realism. Both do the job well enough, but if I had to pick, I think I'd take the Gemini one.
OpenAI is talking up the improved precision in editing with ChatGPT Images 2.5, so I gave it a go after submitting the following prompt: "Create a photorealistic picture of a cabin sat deep in the woods, in its own clearing, with trees all around. It's early morning and the sun shines through the foliage, while the cabin itself looks modern and welcoming."
Click on a ChatGPT image and you can add comments about specific aspects of the image, mark up sections of the picture to change, and quickly erase parts of it or the entire background. I tried getting it to remove some cabin furniture and change the time of day from early morning to dusk, and it completed the brief with aplomb—with the rearranged shadows a particularly clever touch.
Gemini gives you a simpler set of editing tools when you click on an image, but the basics are the same: Use text comments or scribbled annotations to specify the changes. Again, I got the AI to remove some cabin chairs (there must be a lot of chairs in AI model training data), and switch the scene to dusk. Gemini did fine as well, but ultimately the ChatGPT effort looked a little more lived-in and a little less artificial.
These two AI bots are very close in performance when it comes to image generation. Gemini is no slouch, and Nano Banana 2 is excellent at almost everything, but I think ChatGPT Images 2.5 has a slight edge at the moment—particularly if you add in the extra features for sketching and saving prompts as templates.
The debate over when and why to use these tools continues, and I think there will always be more value in actual human art and photography, no matter how good AI gets at making pictures. A real forest is always going to beat an approximation of one stitched together by AI. For visualizations and ideas where you wouldn't employ an artist or have the design chops to do it yourself, though, both Gemini and ChatGPT are at a higher level than ever.
It's the editing tools that are potentially the most useful at this point: Changes to pictures that would usually take professionals hours can now be done by anyone, in seconds, and the capabilities are only going to improve in the future.
Google has rolled out two new AI models: Gemini 3.8 Flash (the "intelligent workhorse model" for everyday prompting), and Gemini 3.8 Flash Cyber (available to "trusted defenders" for cybersecurity protection). While we only got Gemini 3.7 a matter of weeks ago, the new model brings with it some substantial improvements.
According to Google, reasoning and coding are better than ever with Gemini 3.8 Flash, and there are "significant" improvements across "software engineering, agentic tasks, and critical, multi-step reasoning in specialized domains"—so this is very much an update for advanced AI, rather than your basic prompting. You'll find Gemini 3.8 Flash available now inside the Gemini app if you use either the AI Pro plan ($19.99 a month) or the AI Ultra plan ($99.99 or $199.99 a month). There's nothing yet for users on the free tier or the $7.99-per-month AI Plus plan, though these models do trickle down to lower tiers over time, as upgrades are released.
You'd be forgiven for being more than a little confused by all the Gemini AI models Google has pushed out in recent months. For some time now, the models have been split into Flash-Lite, Flash, and Pro, and as you go up that scale, the AI gets slower, more capable at deeper reasoning and complex queries, and more expensive. (Not to mention, these models have different thinking levels too).
With the Pro category, we're still on Gemini 3.1 Pro that launched in February 2026. While a follow-up was mentioned at Google I/O in May, it's yet to be made public. The Flash channel is moving along much more quickly, and it was only a few months ago in December that the first Gemini Flash 3.0 was previewed.
Now we have Gemini 3.8 Flash. As per Google's benchmarks (which you should always take with a pinch of salt), it offers improvements over its predecessor when tackling "long-horizon software engineering" (so full projects rather than small tools), knowledge work (synthesizing and digging into information), complex legal workflows, financial analysis, and agentic activities. Gemini 3.8 Flash is particularly good at complex engineering problems, Google says, and has improved in multi-step reasoning across science, math, humanities, and professional fields. It'll go into more depth to figure answers out, run more checks on their accuracy, and work better at combining different skills in one response.
There are safety measures in place too: The AI has guardrails to block misuse concerning chemical, biological, radiological, nuclear, and cyber offense lines of questioning, and is better than ever at resisting prompt injection attacks (where prompts are engineered to try and get the AI to do something malicious). It all looks impressive on the surface, as do the demos Google showcases here: One of them is a fully working version of Google Maps inside an MS-DOS interface, which will particularly appeal if (like me) you started using computers back in the days of the command prompt.
While I don't have any long-horizon software projects on the go at the moment, I did put the new Gemini 3.8 Flash up against the Gemini 3.6 Flash model that's available to free users in a variety of tasks: designing an interactive Solar System animation, summarizing and simplifying a dense scientific paper, and analyzing the last week of emails in my Gmail account (one of the toughest engineering challenges I know).
To begin with, Gemini 3.8 Flash was noticeably faster: 23 seconds versus 27 seconds for building a space simulation, 15 seconds versus 20 seconds for the science summary, and more or less equal on the emails. Add up those percentages over huge projects and multiple tasks, and you can see that the new model really does get stuff done faster.
For the Solar System test, the simulations were pretty similar—perhaps a sign of how fast they've progressed that a task like this is straightforward for them now. Gemini 3.8 Flash did do a slightly better job at the quality of the renderings though, as the planets look noticeably more "real." When it came to the science summary, again both AI models tackled a tricky problem well. If anything, Gemini 3.6 Flash presented a more understandable summary—perhaps as it's a simpler model—but I found the visualizations and diagrams presented by Gemini 3.8 Flash more helpful (remember the images are created by Nano Banana 2 in both cases, but the info is provided by the underlying model).
On the email challenge, there wasn't much to choose between them: Both accurately summarized what was in my inbox and picked out the same items that needed attention (emails that weren't that important as it happens, but that's a whole other story). This is still a job I prefer to do myself, but these AI models are getting more reliable.
The speed and sophistication gains that Google has talked about are clearly there, even in small-scale tests like this—and it's notable that creating mini web apps and digging through complex physics papers are now straightforward for Flash models as well as the Pro equivalents, which is maybe something to think about if you're considering signing up for a Google AI plan. (Of course, if you want to upgrade to Gemini 3.8 Flash, you'll need to pay.)
AI agents are becoming more capable over time, and the latest ability rolled out to ChatGPT Work (that's the agentic bit of ChatGPT) is being able to sign into websites—and in theory, taking even more of your workload off you. The official examples of how you might want to use this include setting up utilities for a new apartment, organizing an appointment with a doctor, and canceling tickets for a rescheduled trip. "If it's something you’d normally have to open a browser and click through yourself, try asking ChatGPT Work to do it," says the sales pitch.
I'm all for saving time on admin, but handing over my login credentials to ChatGPT feels like a bold step. Can I trust ChatGPT to keep my details safe and secure? And what are the chances of it going haywire once it's in my accounts? In the interests of finding out, I let it log into some of my less important accounts to see what it could do.
This feature is rolling out now in the web and mobile apps for those on ChatGPT Plus, Pro, and Business accounts.
According to the ChatGPT release notes for August 25, 2026, "ChatGPT cannot see your username or password, and they are never seen by the model or used in model training. ChatGPT does not store your username or passwords." So how does it work if your usernames or passwords are never stored? The most basic option here is to simply sign in through ChatGPT when prompted by a site or app (and clear any two-factor authentication challenges at the same time), and then have the AI stay logged in for future tasks.
The more advanced version, which is where you save a more significant chunk of time, is to give ChatGPT permission to interface with a third-party password manager. ChatGPT doesn't specify a list of supported password managers, but as far as I can tell, this is just for mobile: Any password manager on your phone can pop up when ChatGPT needs it and supply the relevant credentials.
Underpinning all this is an actual browser, built into ChatGPT, that it can access and operate itself. Your accounts will see it as a new sign in (like they would if you signed in from a new device), and some websites and apps will let you mark ChatGPT as a "trusted" operator that can log in again in the future,
You can control the data held by this browser: From ChatGPT on the web, click your profile icon (bottom left), then Settings > Cloud browser. You can delete stored cookies from here, for example, which will effectively kick the AI out of any sites it has access to. As per ChatGPT support: "You stay in control of which websites ChatGPT Work can access, and it will always ask for confirmation before consequential actions, such as completing a reservation or payment." Exactly how it can tell when it's about to do something consequential, however, isn't clear.
If you want to make use of this inside ChatGPT on the web, click Work at the top of the screen to switch to agent mode, then tell the AI what you want to do: Something like "sign into my eBay account and tell me how my auctions are doing" or "sign into my accounting app and submit these invoices."
I gave the eBay one a go first—the process proved to be incredibly slow for whatever reason, and I could've signed in multiple times in the normal way by the time ChatGPT got around to gaining access. There are also problems where the ChatGPT conversation tends to obfuscate what's going on: It asked for a verification code, and I didn't know where to check for it.
You can, at any point, switch to the ChatGPT browser itself, which I eventually did to get through the verification code stage (it was sent over text). The AI was then able to correctly tell me what the current auction prices were on the items I was bidding on, after initially stumbling through several wrong pages—there's a thumbnail of the live browser in the corner of the screen, so you can see what's happening.
Next up, I tasked ChatGPT with getting into my Airbnb account and finding some great spots close to Santa Cruz, Calif., within a certain price range. Again, I had to switch to the actual browser view to sign in (I was told the Airbnb site wasn't cooperating), but once I had given the AI access, it worked smoothly enough, looking up properties and giving me a rundown of prices.
At the moment, the sign-in flows are just too finicky. Sites seem suspicious of the AI browser, everything is really slow, and you run into complications with verification codes, passkeys, and the like. You'll often have to manually sign in as you would anyway in order to get ChatGPT up and running on a site.
Once that's done, though, everything works reasonably impressively, as long as you're giving commands that are fairly easy to interpret and carry out. ChatGPT on Airbnb seems competent with dates and locations, but I'm not sure it could filter through places based on the view or the general vibe of the decor.
Perhaps most importantly, I still don't trust the AI to do these jobs better than I can, even if it can (eventually) do it more quickly. I want to see for myself what an Airbnb looks like, or who's bidding on an eBay auction, or what the available time slots are for an appointment to get my car serviced—if I left it to ChatGPT, I'd be too worried about it missing something (I have the same reservations about letting friends and family do things for me as well, but that's another issue).
In my testing, ChatGPT didn't fail in any of the tasks I gave it once it had gained access to these sites, or give me any inaccurate answers, but that would always be at the back of my mind if I switched over to this completely. For now, I'm going to stick to these online admin tasks myself.
We may earn a commission from links on this page.
Part of the fun of upgrading your phone is digging into all the new features and functions it offers, and it's no different with Google's recently unveiled Pixel 11 series. While there are hardware bumps in these phones, many of the most notable selling points are on the software side, with a bunch of upgrades related to AI and Android 17.
If you've invested in a Pixel 11, Pixel 11 Pro, Pixel 11 Pro XL, or Pixel 11 Pro Fold, then these are the best tricks to try out first. All of these hacks are exclusive to the 2026 flagships, though we might see some of them trickle down to older handsets over time (for now, Google isn't saying one way or the other).
HiLight is new to the Pixel 11, and uses a rear LED light to alert you to incoming calls from the most important people in your life—the idea being that it helps you cut out the rest of the noise associated with owning a smartphone in 2026.
You can configure HiLight from Settings by choosing System > HiLight, but there's not actually much configuring to do. It only works for Gemini chats and calls from contacts marked as favorites, with a simple toggle switch in both instances. You can choose different colors for different contacts, but that's about it.
To do more with HiLight, though, you can install HiLight Studio. This third-party tool unlocks HiLight for use with more apps on your phone, and gives you full control over the colors displayed, making HiLight much more useful.
One of the Pixel 11 exclusives in the Camera app is what Google is calling Camera Looks. They work like subtle filters, similar to the Photographic Styles on iOS, with gentle adjustments to contrast and colors that give your images a different feel (and can remove the ultra-clean, over-processed vibe of Pixel photos in general).
Open the Camera and tap the controls button (lower right), then Looks. This gives you access to the four different options you can apply by default when snapping images: Original (the usual Pixel processing), Natural (for "relaxed details"), Shadows (for "deeper contrast"), and Vanilla (for "a warm, golden glow").
Here's a tip about not using a Pixel 11 feature. If you have the Pixel 11 and Pixel 11 Pro Fold you can go up to 30x with Super Zoom, while on the Pixel 11 Pro and the Pixel 11 Pro XL there's Pro Zoom up to an incredible 120x.
The Pro Zoom levels get you closer to the action, sure, but they also use generative AI to enhance what's in the photo. It's not just sharpening and enhancing; it's actually inventing new details based on the pixels that are already there (which is why you have to download a generative AI model the first time you use Pro Zoom).
The 30x Super Zoom uses AI too, but of a different type: Rather than generating details, it's combining photo frames and analyzing colors to sharpen up an image, like a traditional image editor filter. This applies up to a 30x zoom level, Google says, so you know where to stop if you don't want any AI elements being added in.
With Pro Zoom shots, Google Photos will save both the AI-enhanced version and the raw digital zoom version—you can switch between them in the app and choose the one you like best.
It's the perennial problem of the modern age: How do you make sure you've got photo memories of an event while still staying present in the moment? Of course you'd like snaps of your kid's birthday party, but you don't want to spend the whole time glued to your phone.
This is where Magic Capture comes in: You can enable it from the Camera app by tapping the icon to the left of the photo icon, just under the main shutter button. You then tap the shutter button to start a video capture, and then let it record. You can focus on what's happening rather than looking at the Camera app and taking individual photos.
When you stop the recording, AI swings into action to pick out still images that look like highlights from the footage—so presumably someone blowing their candles out, raising a toast, or reaching the crest of a wave, depending on the scenario.
Every phone maker has come up with multiple ways to try and keep us from getting too addicted to our devices, and the latest digital well-being idea from Google is the Pause Point feature for the Pixel 11 phones.
It's pretty simple, which means it just might work: For the apps you specify, you get a mindfulness prompt every time you open them up. Those few seconds of thought and self-reflection could help you realize that no, you don't actually need to open up Instagram or X again today.
To set this up, go to Settings, tap Digital Wellbeing, and then Pause Point to pick the apps you'd like it to apply to. This doesn't seem to be live yet, however, despite Google announcing it and including it in its official support documentation. Keep an eye out for it appearing in updates over the next few weeks.
Another way to reduce screen time is to create unique alerts for each contact, so you know who's getting in touch with you just from the way your phone pings or buzzes. The Pixel 11 phones help with this by letting you customize vibrations in a couple of handy ways.
First, you can set specific vibrations for specific contacts, as well as specific audible ringtones (which has long been an option). From the Contacts app, tap on a name, then Contact ringtone, and you can choose a vibration pattern as well as a sound.
You can also change the default vibration patterns for calls and notifications, which you couldn't before—though not on a per-app or per-contact basis. From Settings, pick Sound and vibration, then Ringtone or Notification.
Attention, content creators! If you're using your Pixel 11 to record video for public consumption, you can make use of a built-in teleprompter to read your scripts—no third-party app required.
From the Camera app, tap the video icon (to the right of the photo icon, under the shutter button), then tap the settings button up in the top left corner. If you switch to the Creator tab, you'll see there's a Teleprompter toggle switch you can turn on.
Once it's enabled, tap the Teleprompter icon on the capture window to enter your text (you can of course paste it in from somewhere else if you want). When the video recording starts, as you begin to talk through the script, it'll automatically advance for you.
This is one of the more well-hidden new features on the Pixel 11 phones. Pixel Search is designed to make it easier to find stuff on your phone (like messages and files) from a single search bar, but it's not the search bar on the home screen.
Instead, it's the one above the app grid. Swipe up on the home screen to get to the app grid, then tap in the search box at the top. The first time you do this on your Pixel 11, you'll see an introduction to the feature, together with a Go to Settings link.
Follow that link, then Apps in search results, and you can choose the content that Pixel Search will dive into—across Google Calendar, Google Drive, Gmail, Google Photos, and Settings, for example.
There's an exclusive Gboard add-on for Pixel 11 devices called Rambler, which Google has positioned as a super-smart upgrade on voice dictation. Phones have been able to turn speech into text for a long while now, but Rambler is better at dealing with natural speech rhythms, filler words, and pauses.
It's also better at editing. If you say "the meeting at 2pm... or rather 3pm" in your ramble, the "3pm" will make the finished transcript. If you're reviewing text Rambler has transcribed, you can launch Rambler again to make edits to it with natural prompts, too—something like "change the day from Friday to Saturday."
Load up Gboard on your Pixel 11 (any text input box will do), tap the microphone button, then read through the Rambler introduction and tap Get started. Once it's set up, it can be launched at any time through the microphone button.
You can go back to the standard voice dictation at any point, if you prefer it: From Gboard tap the menu button (the four squares on the left), then pick Settings > Voice typing.
You can use StandBy mode to turn an iPhone into a bedside clock, and the Pixel 11 handsets have something similar—sort of. From Settings, choose Display and touch > Screen saver, and you'll see there's a new Charge option there.
This displays a progress bar showing charge status, and a clock. Enable it, and if you hook your Pixel up to a magnetic wireless charger overnight, you've got a decent, minimal bedside display that works in portrait or landscape orientations.
Installing the ChatGPT desktop app gives you a few extras that aren't available through the web portal, including the ability to run prompts that query your local files, emails, and screenshots. Now, with the macOS client, those additional features extend to your iMessages, too: There's now an Apple Messages plug-in available.
The idea of giving ChatGPT access to your chats is something a lot of users aren't ever going to countenance—"good luck with that spy software" is one typical response on Reddit—but we know that plenty of users are already asking AI for help in composing and interpreting messages. This just makes that process more streamlined. (Claude has an Apple Messages plug-in too, for what it's worth.)
As OpenAI confirmed to TechCrunch, the plug-in does all its processing on your Mac, and doesn't send anything about your chats back to servers in the cloud—these Messages conversations won't appear in your ChatGPT apps on the web or on mobile. In addition, the ChatGPT desktop app doesn't build a full index of your text messages.
This is something that's available for all ChatGPT users who have a Mac: To get started, download and install the desktop app for macOS, then open the Plugins tab, search for "messages," and click Install next to the Apple Messages entry. You can then use natural language in your prompts to invoke the plug-in ("send a message to..."), use the @Messages mention, or select Messages from the plug-in list in the prompt box (click the + icon to find it).
Once you've set it up, ChatGPT can then do just about anything you can imagine with your messages, starting with the basics: It'll send messages for you, for example, and even compose them if you're stuck for inspiration. You just have to name the contact, so something like "message Steve telling him I'm sorry I can't come to his dinner party in a polite way" will do exactly that.
When I tried this, ChatGPT whipped up a suitably friendly message, with the right blend of thank yous, sorrys, and hopes for a lovely evening. When sending messages, you get the option to review what's being sent and approve it before it goes out—you can approve messages each time, or have them sent automatically in the conversations you select.
Then there's everything the AI is able to do with your message history. You can use prompts like "who's messaged me the most this week?" or "is there anything I need to follow up on in my conversations?" and have ChatGPT take a look.
I tried getting ChatGPT to dig back into my message history for details of event arrangements and deliveries, and it picked out the right information each time. I also got it to look back over text messages from my phone carrier to see if my bill had gone up over time, which it had: The AI gave me the exact increase, and the percentage increase, together with the relevant dates.
Conversations can continue based on whatever you find as well, so if you discover there's a chat thread with an invite that you haven't responded to, ChatGPT can sort that for you—and may even draft a suitable reply you can approve with a couple of clicks, depending on the context.
It's in this message history retrieval and analysis where the ChatGPT plug-in seems most useful to me. I'm reluctant to start farming out message composing and sending duties to AI—it seems rude, first and foremost, not to mention bad for my brain—but searching through reams of text is something that AI does well, and could genuinely end up saving me some time.
If you're prepared to trust ChatGPT with access to your Apple Messages chat list (and that's a big "if") then it's another database for the AI to mine and analyze. As with a lot of these other plug-ins, like Claude and Gmail, it's a question of where you're going to draw the line between privacy and convenience.
And there are certainly risks to watch out for: I can see this being yet another way that people become over-reliant on AI and its "thoughts." If you want ChatGPT to assess someone's mood over time based on their chat history, or compose a message that gives you the best chance of rekindling a relationship, or write a text getting you out of work based on a fake excuse that sounds genuine, ChatGPT will do it. That's where the tech veers beyond the useful and straight into the uncomfortable.
If AI is going to take all of our jobs, the least it can do is make those jobs a little easier while we still have them, and email admin is one of the biggest time sinks I've got in my daily routine. I'm spending hours a week reading and responding to emails when I really need that time to get on with writing about new Pixels or Mac apps.
My interest was piqued then by the news that the Claude AI chatbot can now respond to and send emails in a connected Gmail account–and without your specific approval, if you're happy to let it work automatically.
Claude has been able to plug into Gmail accounts for a while, so you could already get it to read through and summarize incoming messages—but any email drafting had to be done in the Claude bot and copied over. Now the AI is able to compose and send responses inside Gmail without you having to switch over to the Google email client yourself.
Could this free up some more time in my day? Is it really wise to hand over emailing duties to an AI bot? I gave it a try to assess how it might impact my productivity (and sanity).
You need a paid Claude plan for this at the moment, so you're going to have to be on the $20-a-month Pro plan or something higher. To connect your Gmail account if you haven't already, open Claude on the web and click your profile picture (bottom left), then select Settings > Connectors and find Gmail. (If you don't see your profile picture, click the menu button, top left.)
To do something in Gmail, you just need to mention it in a prompt. If you're not quite ready to do some drafting and sending, try a prompt like "who's emailed me most today" or "summarize the emails I've received this week." Claude rummages through your Gmail account and delivers what you asked for as a response.
When you're ready to start sending, you can be pretty informal and ambiguous with your prompt. Saying "reply to my last unread email," "reply to David Nield," and "reply to the Story edits thread" will all work in Claude. However, the more precise you can be—with a specific email address, for instance—the better, in terms of avoiding mistakes.
If you say "send an email to Dave" and Claude finds three people with that name, it will ask who you mean. However, if it finds the wrong Dave and misses the other two for whatever reason (maybe due to formatting or email address construction), it will make the incorrect pick and not tell you about it. So, be as precise as you can.
Next, you tell Claude what to say: You can dictate the message word for word, or tell Claude to get creative and use its imagination, or do a mix of the two. This works the same as any other text prompt—supply as much or as little direction as you want regarding email content, style, intention, and other characteristics.
You'll be relieved to hear that automatic sending is off by default. If you go back to Settings > Connectors, you can click Gmail to set permissions. Different actions, like searching through your inbox, sending emails, and editing labels, can be set to one of three different options: Never allow, always allow, or always ask for permission.
I took the risk and gave Claude permission to send emails automatically without approving each draft and got to work. I did consider getting in touch with the Lifehacker editors, but keen to stay on the roster here and avoid any hallucinated misunderstandings, I decided to have an email conversation with myself, via a secondary email address.
Claude actually works really well for getting around your Gmail inbox: It understands labels, dates, contacts, subject headers, themes, and nuance. It easily identified "the email I've just sent myself from my Proton account" without me having to open up Gmail, and I told it to reply—saying that it should agree to a meeting and talk about the weather.
One immediate problem: It signed off my email with "cheers,, something I would never say outside of specific scenarios involving bars and clinking glasses. I don't think I've ever used that in an email, but evidently a lot of people in Claude's training data have, which is why it was included. Be aware that Claude may not catch your default tone and personal nuances when you let it compose your messages.
The rest of the email was fine, if a little generic: I agreed to a meeting (with myself), and mentioned the weather (a standard British conversation topic). I then repeated the process several times, giving Claude different levels of freedom when it came to composing message replies.
On a technical level, it's all quite impressive. Claude can accurately pick up and transfer what's in an email thread and compose replies as instructed. I told it to generate a message summarizing my work for the week (via its Google Drive access), and to add a question about a weekend soccer match, and it did both with aplomb. We know AI makes mistakes, but I didn't see any in my testing.
As clever as this is, though, the biggest issue is that I can't really see myself using it. To begin with, these emails aren't written in my voice or style, and I'd be continually worried that Claude was about to make a terrible error somewhere (maybe in misinterpreting an incoming email, or getting something wrong in an outgoing one).
Another concern is where all this is heading: AIs emailing other AIs, into infinity? I already suspect a lot of the emails I get are crafted by chatbots, which immediately devalues them. If I'm going to communicate with someone, I feel I owe it to them to actually put together the message myself. While I think I'll keep Claude's Gmail connector around to search and summarize emails, the sending and drafting features are getting turned off for now.
There are now enough freely available Large Language Models (LLMs) that running them locally on a computer is commonplace, and a better idea in some ways—they're more private, as you don't have to send anything to the cloud, and they work offline.
Getting these advanced AI models running on phones hasn't been quite as straightforward (hence why the most advanced Siri AI features need one of the latest iPhones), but we're now at the point where most of the handsets are powerful enough, and some of the models are small and efficient enough that it's actually practically feasible.
The benefits are the same as on the desktop—you get an LLM that's always available and private to you, with nothing sent back to Google, OpenAI, Anthropic, or anyone else. The downsides are that you're going to get slower and more limited performance from your AI, as you're dealing with smaller and less powerful models, and you might see a hit with battery life (these AI chats can be quite resource-intensive).
Even with those compromises, the local AI you'll be able to get up and running on your phone is still going to be more than capable of everyday tasks and chats—and that means you can give Gemini or Siri (and perhaps your AI subscription) a rest.
Any iPhone or Android phone launched in the last couple of years should do a decent job of running a local LLM; the AI boom means manufacturers have started building their devices with this kind of usage specifically in mind. Older phones can still apply, but you might need smaller models and get worse performance.
RAM is actually a bigger consideration than chipset performance when it comes to running an AI model. You're going to need 6GB or above for this to work at a satisfactory level, and 8GB or more is better for the larger models. Below 8GB, stick to 1-2B models—that B is for billion, and refers to the number of parameters in a model (essentially, how smart and versatile it's going to be).
In terms of storage space, you don't have too much to worry about. Even the largest models that use 7-8B are going to top out at around 5GB for how much room they take up on your phone, so you might even want to keep several AI models on your handset and switch between them as required.
If you buy a new flagship-level iPhone or Android phone today, it will actually come with local AI models preinstalled for quick tasks that can be done without pinging the cloud. However, these Apple and Google-made models aren't available directly to you, the user—they're used as and when Siri AI and Gemini think they're the best option.
There are several apps that will do the job of getting AI models on your phone and running them. A couple of the most popular are Atomic Chat (Android or iOS) and PocketPal AI (Android or iOS). There are some differences—it's easier to extend Atomic Chat to desktop LLMs, for example, while PocketPal AI is a little more lightweight—but they're both great for getting started with local AI on your phone.
As for models themselves—sometimes called SLMs or Small Language Models because they're designed to work in more limited environments—you've got plenty of choice here, too. For instance, Gemma is the umbrella name of the open-source models that Google makes available for free, and some of these are specifically engineered for working in tighter spaces with fewer resources (like on your phone).
Meta has its own open-source AI models given the Llama moniker, while Microsoft has its Phi-4 models, which are also highly rated for efficiency. You just need to look for the versions with the lowest number of parameters in front of the "B" (or with "mini" in the name) to find the packages that'll work best with your phone.
At the moment, these SLMs are mostly text-only, although some of the newer, larger, and more advanced ones can analyze images and files. If you want to be able to generate images and videos, you're going to have to use the conventional cloud-based AI models, at least until the next leap forward in the technology.
To see how useful one of these local AI models might be on a phone, I installed PocketPal AI and one of the smaller Google Gemma models on my Pixel 9 Pro—not really a flagship any more, but not too old. I did like the little selection wizard that PocketPal AI opens with, that directs you straight to a suitable AI model for your phone.
Once I'd downloaded and installed a couple of models, it was simple enough to load them up and get started. PocketPal AI also gives you access to optional "pals" (hence the app name) that can tailor AI models for your use—the default Pip option seems fine as a stand-in for what you might be used to with the standard Gemini or Siri AI apps. Prompting and following-up works as normal, with your chat history saved by default.
There is a noticeable (and expected) slowness to the responses when you run LLMs on your phone, and a noticeable difference between AI model sizes—choosing a smaller model will get you answers significantly faster, even if they're not quite as smart or complete. It's worth experimenting with a few models just to find your own sweet spot between performance and speed.
With no web search or up-to-date knowledge available, this is best for brainstorming ideas, analyzing and refining existing text, composing new text, and getting fast facts or comparisons ("give me a film like..."). As always, watch out for those hallucinations, and don't take the word of any AI to be guaranteed as accurate.
If you've read our previous coverage on the Tor network and the Tor Browser, you'll know they're the best choices if you want to maximize your privacy and anonymity on the internet. Now, the Tor project has teamed up with the Bloco Android app studio to release another app focused on privacy and anonymity: Snowflake Volunteer.
This isn't only bout keeping your own browsing safe and secure—if you install it on your Android phone, you're helping other people circumvent censorship restrictions in their current location, donating part of your connection to the web so that they can get online at all.
The Tor Browser is so effective at hiding the identity and online activities of users that certain governments would rather their citizens didn't make use of it. That takes away one of the best privacy browsers in the business for the people of those countries, and anyone who happens to be visiting them. Censoring the Tor Browser is usually done by refusing access to the relay nodes that it routes traffic around—the stops on the way to and from the internet that do the job of concealing locations and other identifiable information. However, the relay directory is public, which means it can be blocked.
That's where Snowflake comes in. The standalone Android app is new, but the idea isn't. Snowflake is an alternative way the Tor Browser can get online: A Snowflake connection is disguised as a standard video call over the WebRTC protocol, which isn't typically restricted, and from there, a connection can be made to the Tor network.
With Snowflake on your phone, you're essentially smuggling someone in a Tor-restricted country across the border, digitally speaking. Even if these intermediate bridges are detected and blocked, they're short-lived, with new ones continually popping up, making it virtually impossible for a government to block this method of access without crippling much of the general internet in their country.
The Tor Browser runs as normal for the user who's connecting, albeit with the standard trade-offs in terms of speed and latency. Because of the rerouting and technical borrowing that goes on, Tor isn't the best option if you want to maximize your broadband upload and download speeds.
As a firm believer that internet users should be able to get on the web with a high level of anonymity and privacy, I installed the new Snowflake Volunteer app from the Google Play Store (it's also on F-Droid) and powered it up. The first job is to exempt it from battery management rules so it will stay available even if it's not being used, a process the app guides you through.
That done, all that's required is enabling the toggle switch on the main screen, and your phone is ready and waiting for someone else to connect. There's no setup configuration to work through, no user account to create, and everything works automatically in the background—it really is designed to be as simple as possible.
Tap on the settings cog (top right), and you'll find a few settings at your disposal: You can opt to only allow connections when you're on unmetered networks (such as wifi) or when your phone is charging. You can also set how many users your phone can support at any one time. This is all in the interest of limiting the impact on your device.
Across four hours or so, I saw six connections made, sending a total of 3.2 megabytes and receiving a total of 17 megabytes—hardly excessive use. My phone wasn't suddenly overrun with other users wanting to get online. Snowflake Volunteer is designed as an emergency measure for circumventing censorship, and the usage I've seen backs that up.
I didn't notice any bugs or any problems with the app or any impact on my battery life (perhaps to be expected with so little activity). I think it's something I'll keep switched on, especially overnight when I'm not using my device. It's a nice idea to be able to see the stats on how many people you've helped over time, as an encouragement to keep your device available.
As a Snowflake Volunteer, you're not at risk of being identified by any state that's censored the Tor project: To them, you look like a standard audio or video call connection, and there's no static list of IP addresses to block. What's more, the same multiple layers of encryption that protect the entire Tor network are still in place, so you're unable to see anything connecting users are doing, and they will go through several more nodes before doing any browsing anyway—your device is just the entry point. No one who connects to your device can learn anything about you: The app is simply not capable of any sort of surveillance. You're loaning a slice of connectivity, anonymously and securely. (There are more details of how it all works in Tor's Snowflake documentation.)
The new Android app isn't your only option for putting your hand up as a Snowflake volunteer. You can also install the desktop browser add-on or enable the Kindness Mode in the Orbot app for iOS (a VPN that enables connections to the Tor network).
We may earn a commission from links on this page.
If your time on macOS is solely spent switching between the well-known apps that everyone uses—your web browser, Notes, Apple Music—then you're missing out. There are a host of more unusual and idiosyncratic apps for Apple's desktop operating system that are well worth digging into.
The apps collected in the list below fall under that description, and while they might not have been installed on millions of Macs or be particularly complex in what they offer, they're all quite useful. Give them a try and see how they can upgrade your macOS experience.
Scrolling isn't particularly hard on macOS, but what if it were even easier? ScrollPods teams up with a connected pair of AirPods to let you scroll up and down webpages, documents, and more with a tilt of the head. It works with anything on screen that can be scrolled.
The utility is simple to set up, and based on my testing, it works quite well. The default sensitivity is just right, with slight head tilts enough to nudge a page up or down, but you can adjust the sensitivity and set up custom gestures for the app through its settings.
ScrollPods (free)
You can add some extra functionality to your MacBook trackpad with Slidr: It turns the left side of your trackpad into a brightness slider, and the right side into a volume slider, so you can quickly adjust these key settings without interrupting your typing and scrolling flow.
You're able to take advantage of a free trial before committing to the purchase, so you can see if it works for you. It couldn't be any more straightforward to use: The app runs from the menu bar, and there are options for adjusting the sensitivity and position of the slide gestures, and swapping over the brightness and volume sliders if you prefer.
Slidr (free trial, then $4.99)
A picture-in-picture mode can be very useful when multi-tasking, but it's not supported in every app—something Pipiri fixes. Once you've got Pipiri installed, any window on the macOS desktop can be put into a picture-in-picture mode, running on top of your other apps.
So, for example, you can keep an AI agent window or a social media feed visible on screen while you get on with something else. There are lots of options to play around with, too, including the refresh rate of the overlaid window, app hotkeys, and adaptive resizing.
Pipiri (free trial, then $9.70)
Haptyk takes advantage of the hidden accelerometer that's built into every Apple Silicon MacBook keyboard, figuring out how hard you're striking each key and adding appropriate audio. You get mechanical keyboard sound effects, with your existing MacBook keyboard.
I gave the app a go and it works really well: There's zero lag on the sound effects, and you get the same experience across every app. You've got 16 different keyboard sound packs to choose from, plus lots of configuration options (including volume, pitch, and gain).
Haptyk (free trial, then $8)
If you're a regular video caller, then you'll know that subtle feeling of dread that rises up whenever you're called upon to share your screen or to record something you're doing. What browser tabs do you have open? Is an embarrassing notification going to pop up?
Stealthy deals with this with the minimum of fuss, hiding your desktop icons, the desktop wallpaper, and the dock. silencing notifications, and clearing away open app windows you don't want visible on screen. It can also launch automatically when video calls start.
Stealthy (free trial, then $12.99)
There's nothing too revolutionary about Gestimer, but it does a simple job in an intuitive way, and is likely to come in handy more often than you might think. It lets you set timers and reminders by just dragging down the app icon from the macOS menu bar.
There's no messing around with opening up other apps or digging through calendars—it's just click and drag, with the length of the drag controlling the time that's set. There's a persistent countdown in the menu bar, and you get a notification at the appropriate time.
Gestimer (free trial, then $6.99)
Unclack is both simple and useful: It mutes your microphone when you're typing (even more helpful if you've installed Haptyk, above). That means you no longer need to worry about your aggressive typing style distracting other people during video meetings.
It runs quietly from the menu bar and you don't need to do anything to enable or disable it—it just works, and you won't even know it's there. Should you need to talk and type at the same time, you can disable Unclack with a couple of clicks until you need it again.
Unclack (free)
Almost a full year has passed since the launch of Google's Pixel 10 phones, which of course means it's time for the Pixel 11 handsets to make their debut. Another Made by Google event has been scheduled for Wednesday, Aug. 12, with everything getting underway in New York from 6 pm EDT / 3 pm PDT.
As Google usually launches new Pixels at this time of year—and has already dropped some pretty revealing teasers on the Google Store—we have a good idea of what's coming. Add in a recent flurry of leaks and rumors from reputable tipsters, and it's possible to generate a clear picture of what will be announced next week.
Starting with the Pixel 11 phones, it looks like there will be four new models, just like last year: The Google Pixel 11, the higher-spec Pixel 11 Pro, the higher-spec and larger Pixel 11 Pro XL, and the foldable Google Pixel 11 Pro Fold. Google has already namechecked and teased the design of the Pixel 11 Pro and the Pixel 11 Pro Fold on social media.
You can see from those official teasers that the familiar Pixel design language is back, with the addition of a colorful notification LED on the rear camera bar. Early rumors suggested this would be called "Pixel Glow," but reputable tipster Roland Quandt has dug up some material suggesting it'll actually be called "HiLight," and will be used to indicate Gemini chats or incoming communications from priority contacts.
This Tweet is currently unavailable. It might be loading or has been removed.
For everything else, we're relying on unofficial sources. Veteran leaker Evan Blass, who has one of the best track records in the business, has shared renderings of all four expected phones, including the Pixel 11 and the Pixel 11 Pro XL that Google hasn't specifically mentioned so far.
Aside from HiLight, there aren't really any design surprises here, but Google is apparently once again tweaking the color options. A comprehensive leak published at MyMobiles lists Blue, Green, Midnight, and Pink for the standard Pixel 11; and Green, Light Green, Silver, Midnight, and Red for the Pro and Pro XL.
In terms of specs, most reports suggest the devices will have a Tensor G6 chipset, up from the Tensor G5 in the Pixel 10. Speed gains might not be huge, but it will be the first Pixel processor to be based on a 2-nanometer manufacturing process, which should make it more efficient. A new modem is also expected, which should provide faster and more reliable 5G.
This Tweet is currently unavailable. It might be loading or has been removed.
A full specs list published by Android Headlines shows 12GB or 16GB as the RAM options, and storage starting at 256GB (rather than 128GB, as was the case last year). That extra storage comes with a price bump: These phones are predicted to start at $899 for the Pixel 11, $1,099 for the Pixel 11 Pro, $1,299 for the Pixel 11 Pro XL, and $1,899 for the Pixel 11 Pro Fold, all $100 more than their equivalent Pixel 10 models.
As for cameras, the rumors suggest they will stay largely the same as last year, which means triple-lens cameras on the back of every model. Battery capacities are expected to be more or less the same across the board too, with slight increases or decreases depending on the model.
Phones aren't the only new devices tipped to show up on Aug. 12. New Pixel handsets usually arrive alongside a new Pixel Watch as well, and that looks certain to be the case this year. While Google hasn't officially mentioned a smartwatch is coming, seasoned tipster Steve McFly has revealed a host of renderings for the Pixel Watch 5.
Once again, these renders show 41 mm and 45 mm variants in the standard Pixel Watch design, while slight battery capacity increases are rumored. However, the chip inside (the Snapdragon Wear W5 Gen 2 SoC) is believed to be the same as the Google Pixel Watch 4, with pricing is said to start at $399 for the 41 mm version and $429 for the 45 mm version (up by $50 and $70, respectively, over the Pixel Watch 4 models).
This Tweet is currently unavailable. It might be loading or has been removed.
As an added bonus, it's looking very much like Google will finally launch its own AirTag rival at this event. Images and pricing have emerged for the Google Pixel Tag, which looks like a fairly nondescript little Bluetooth tracker, and which will no doubt work through Google's Find My portal to help you recover lost gadgets.
The leaks so far haven't given any hints about a rechargeable battery or support for the ultra-wideband (UWB) tech that makes AirTags so accurate, but European pricing is expected to be €34.90 for a single Pixel Tag. That's almost exactly the price of an Apple AirTag in the region, so a $29 price in the U.S. seems likely.
We may earn a commission from links on this page.
If you own an Android phone, it's more likely to be a Samsung Galaxy handset than anything else—Samsung is comfortably ahead of the likes of Google, Xiaomi, and others when it comes to global market share. And if you've got a Galaxy, you've got access to a selection of features you won't find on other Android devices.
Samsung has added a bunch of its own ideas and settings on top of Google's mobile operating system, and if you want to make the most of your Galaxy, you need to know about them. Here are some of the best ways to hack your Samsung Galaxy experience.
Why keep changing the settings on your phone if your phone can do it for you? Routines is a Galaxy utility that will wait for certain triggers and then take certain actions—one example would be turning on Do Not Disturb mode whenever you reach the office, then turning it off again when you leave.
Head to Settings then pick Modes and Routines. The first screen is Modes—specific notification settings for sleeping, working out, gaming, and so on—but if you switch to the Routines tab you can start adding some new automations.
The available triggers include times, places, specific wifi networks (like home or the office), connected devices (such as headphones), and events (like incoming notifications). Actions to take cover everything from specific Do Not Disturb modes and display settings to tasks inside apps (such as creating calendar events) and accessibility options.
If you haven't explored the Edge Panel feature much, you might not have realized just how capable it is. It's not just for shortcuts to apps—you can also add shortcuts to your favorite contacts, which is potentially even more useful.
From Settings, select Display > Edge Panels, then activate the feature via the toggle switch and tap Panels. Make sure the People panel is selected, and use the Edit button to select specific contacts. Tap the three dots then Reorder to put the contacts panel first.
There are plenty more Edge Panel customization options on these screens as well. You can choose which side of the screen it shows up on and how transparent it is, for example.
Samsung Galaxy phones can take one of your photos, apply it as the wallpaper on your device, and then dynamically adjust it based on the time of day and weather conditions—dimming the backdrop and covering it with raindrops if there's a storm at night, for instance.
You can configure this from Settings by tapping Wallpaper and style, then Change wallpapers > Create with AI > Weather. You'll need to pick an image and allow location access, and the new wallpaper will be applied to the lock screen. To make further changes to the lock screen, press and hold on it while your handset is locked. You're able to then adjust elements like the shortcuts and widgets on the display, as well as the wallpaper.
Galaxy phones come with a special brightness and vibrancy boost for watching video content, whether it's something you've saved locally or something you're streaming. From Settings, tap Advanced features > Video brightness, select Bright, and then use the toggle switches to select the apps it applies to.
Besides beeps and vibrations, notifications on Galaxy handsets can have some cool lighting effects associated with them as well. It's a fun way to make sure you see your incoming alerts, and it also can give you a quick idea of what the notifications are for.
Open Settings, then select Notifications > Notification pop-up style > Add lighting effect. You've got a choice of styles and colors to choose from, and you can set the duration for the lighting effect. For the lighting effects where you can choose a color, you can change that color based on matching keywords by selecting Lighting effect color by keyword. You could put in a contact's name and assign them a special color, for example.
This one is simple but useful: You can change what a double tap on the side button does by opening Settings and choosing Advanced features > Side button. You can have it launch the camera, the flashlight, or a specific app, for example. On the same screen, you can also adjust what a long press on the side button does—you can have it either launch one of the digital assistants installed on the phone (like Bixby or Gemini) or bring up the power menu.
There are likely to be multiple situations where you're trying to use your Samsung phone with one hand rather than two (walking the dog, carrying groceries), and that's where one-handed mode comes in.
It's enabled from Settings via Advanced features > One-handed mode, and you can then activate it by swiping down at the bottom of the screen to get a smaller, more reachable interface. Repeat the gesture again to go back to normal.
Multi Windows is another useful feature available on Samsung Galaxy phones—though it's likely to be most useful on the biggest phone displays. It lets you get two apps on screen at the same time, so you can read messages while watching a video, or whatever you want.
Swipe up from the bottom of the screen to the center to bring up the carousel of recently used apps, then swipe to your first app. Tap the app icon at the top, then Open in split screen view. You'll then be asked to choose another app to pair with it.
One of the lesser-known options on Samsung Galaxy phones is the ability to change the pixel resolution of the screen—a higher resolution means sharper text and graphics, though it will also drain your phone battery faster too, so bear that in mind.
You can configure this by selecting Display > Screen resolution from Settings. The options you'll see will depend on the specific model of handset you're using, but above the options you can see a preview of the difference each one will make.
It's a little odd that Samsung has hidden this menu away, but it's a suite of wifi diagnostics tools that you can use to check signal strength around your home, troubleshoot network problems, and check data usage. To enable the menu, which is called Connectivity Labs, open Settings and pick Connections > Wi-Fi. Tap the three dots (top right), then choose Intelligent Wi-Fi, and tap Intelligent Wi-Fi at the bottom of the screen seven times to add the extra menu.
Open Connectivity Labs and you'll see there are a host of options and features to work through. One of the most useful is the Home Wi-Fi inspection feature: Select this and your phone will guide you in measuring wifi signal strength around your home.
Known by the codename Nano Banana 2, the most recent AI image generation updates added to Google Gemini have been a significant step forward in terms of quality and accuracy, and now similar capabilities are being plugged into the Google Earth app on the web. It means you're able to use the aerial, satellite, and 3D imagery packed into Google Earth as the basis for prompts for AI images—maybe peering into the distant past with a historical reconstruction or looking to the future with a sci-fi reimagining.
We've already seen just how precise and realistic the Nano Banana-powered features in Google Gemini are now, and Google has five suggestions for how you might deploy these capabilities in Google Earth. First, you can learn while exploring, so maybe getting the AI to build an infographic on the Statue of Liberty with key historical facts about it.
A text prompt such as "reimagine this empty lot in Tokyo as a vibrant shopping and retail district with open spaces" would transform a rather bare part of the Google Earth 3D map into something quite different. You can also visualize personal projects such as a dream lakeside home or an extension for your current property. Google Earth would combine the existing landscape with whatever you prompted, and help you come up with ideas as part of the planning stage.
You can also look forwards and backwards. "Sometimes it’s fun to let your imagination run wild," says Google, suggesting that you might want to get the AI to imagine how a place might look in a hundred years. (It might be worth nudging your prompt towards a utopian rather than dystopian vision of the future.)
Finally, Google suggests you dive back into the past. The company showcased a demo with the prompt "render a hyper-realistic view of what these Pompeii ruins looked like in 78 AD," which could be used by teachers, for example. That would instantly transform today's ruins "into a bustling, colorful street scene from the Roman Empire," in Google's words.
These new features come with the usual AI image caveats, however. While Nano Banana 2 can pull real-world information and images from the web, and is able to render graphics with more consistency and accuracy than previous models, there's still going to be the usual weaknesses involved with AI: Your lakeside home might not be entirely structurally sound, and the Pompeii reconstruction may not be entirely authentic—so bear that in mind when your images are produced.
When I asked Google about this, a representative emphasized the real-world knowledge base that Nano Banana 2 draws on, which includes all the text and graphics it can find online, as well as the vast number of images that have been plugged into the AI during its training. With Google Earth, it will also trace over the structure ruins that are still in place when it comes to Pompeii, for example.
For a bit more clarity, you can try quizzing the Nano Banana 2 model inside the main Gemini app: I got it to render an imagined scene from Pompeii just prior to it being smothered by Vesuvius, then asked about the image accuracy. The AI said it had combined a bunch of real, historical information points to render the scene, including details on Pompeii's raised sidewalks, and the clothing styles of the day. Google hasn't said anything specific about usage limits, but the Nano Banana 2 upgrade is rolling out now globally, for all users of Google Earth on the web.
Finding yourself locked out of your Google account for whatever reason is a stressful situation to be in—what with all your emails, documents, photos, and other data suddenly out of bounds—but there's now a new way to prove your identity and regain access: a simple selfie video.
It's always wise to have multiple sign-in methods set up, beyond the basic password and email address, as each additional one gives you a better chance of getting your account back if something goes wrong. The recovery contacts feature is a good example, where Google uses verified, trusted contacts of yours to prove your identity.
Now you can also use a selfie video. By following this link you can see if the feature is live for your account, and record your selfie via a webcam or phone camera. If you're having trouble logging in, you'll be able to record a new selfie, which will be compared against the original saved in your account.
At the moment, the selfie video option for my account isn't available, so I couldn't test it out. It seems this will be rolled out gradually to users, though Google hasn't said anything specific about availability or a schedule—just that users can now check if their account is eligible.
"Your selfie video is yours and you're in control," says Google in its announcement blog post. "It is recorded and securely stored with your consent, and you can delete it at any time in your Google Account. It's used only for helping you sign in, unless you opt to share it for additional purposes. Your selfie video is encrypted at rest, meaning it's securely stored even when it's not being used."
Those "additional purposes" mean that Google will use your "video and related data to help ongoing efforts to develop and improve facial recognition, age estimation, and other verification methods that may use your physical features or movement" (the specter of age verification looms large again).
There's also the rather obvious issue of deepfakes. Google says there are "multiple layers of security" to guard against fake photos and videos, including getting you to go through several simple movements when recording the selfie—movements that are hard to predict and mimic with a deepfake.
All of the usual Google account checks will be in force too, so (for example) you might find additional security questions pop up if you're suddenly trying to log in from the other side of the world from where your usual login attempts are.
Given the accuracy of technologies like Face ID these days, video selfie sign-in should prove to be a useful backup method of accessing your account. There's nothing extra for you to remember—you just have to bring your face—and it appears at the moment to be difficult to get around.
Amid another busy Galaxy Unpacked event and the reveal of several new Samsung foldables and smartwatches, there's also news from Google in partnership with Samsung: Several new Android features are going to show up first on the Z Fold 8, Z Fold 8 Ultra, and Z Flip 8, before appearing on Pixels and other phones.
This has become something of a tradition with Samsung's mid-year product launches, as a smattering of new features—including gaming support for Circle to Search—appeared first on the Galaxy Z Fold 7 and Galaxy Z Flip 7 last year.
For 2026, the emphasis is on Gemini. First, the Gemini Intelligence feature that completes tasks for you, launched in beta in May, is being expanded to support more than 40 apps and is available out of the box on the new Galaxy foldables. Plus, you can launch your agentic assistant with a screengrab now as well as a prompt (so maybe a picture of an item of clothing you want the AI to look for online).
Google hasn't specified which 40 apps are now included, but the idea is the same as before: Gemini can carry out more admin for you, like booking cabs or reserving tables, in your apps. You'll still get an overview of what the AI is up to on your screen, as well as final approval on any important actions, and you can stop the process at any time.
Next up is the newly rebranded Gemini Notebook (the AI research tool previously known as NotebookLM). There are no new features here, but the app is going to be preinstalled on the foldables that Samsung just unveiled, and on the Z Fold 8 devices, you can take advantage of the wider screen to drag and drop elements (such as images and web links) into your notebooks from other apps.
Additionally, if you buy a Galaxy Z Fold 8, Galaxy Z Fold 8 Ultra, or Galaxy Z Flip 8, you get six months of Google AI Pro for free (usually $19.99 a month). Compared to the free tier, Google AI Pro gets you video generation, the Antigravity coding tool, YouTube Premium Lite, 5TB of storage to play with, and significantly higher usage limits across all of Google's AI tools (including Gemini and Gemini Notebook).
The third Gemini perk for those investing in a new Samsung device applies to the Galaxy Watch 9. On the new smartwatch, Gemini will start listening for instructions as soon as you raise your wrist to talk—no wake word required. It's a small tweak, but means you'll get faster access to the AI.
Lastly, Google has also announced "streamlined switching" from iPhones, built into Android 17 and available on the new Samsung phones. You'll no longer need a cable, as it now works wirelessly, and extra data types are now supported as well—including your Google Account details and wifi credentials, on top of photos, videos, contacts, messages, and calendars.
This improved switching method has already started rolling out on Pixel devices, Google says, and will, reportedly, soon be showing up on other Android devices as well.
While Apple has recently been forced to offer alternative app stores in the EU, they've been around much longer on Android: The more open nature of Google's mobile operating system means it's easier to access apps that haven't been specifically Play Store approved, and the quality and breadth of these apps is better than you might have realized.
Leading the way here is F-Droid, a repository of Android apps released as free and open source software, with an emphasis on security and privacy—both the code and the permissions of the apps are presented transparently, and there aren't the usual Google tracking and analysis extras on top.
You don't get the full Play Store security protection with F-Droid, and you are relying on the F-Droid team keeping its operations safe and secure—something it has a strong record in. However, down the years this alternative app store has proved a reliable source of safe and trustworthy apps, including the recommended ones below—just be sure to stick to apps that are well-known and frequently updated.
This is a timely moment to be recommending F-Droid apps, because the database is under threat from Google's stricter policies on apps from outside the Play Store, coming into force in several months time. It's going to mean more verification and less privacy for developers, and a more convoluted process for sideloading apps (these will still be allowed, but you'll have to wait 24 hours for each install).
F-Droid has put up a campaign page you can read through to understand the issues at stake, and you can sign your support to help it battle this "existential threat." In the meantime, the best way to support F-Droid and its openness is by making use of the platform, and the excellent apps below.
Head to the F-Droid site to get the app store up and running, which then gives you access to the app library, including those here. Once these apps are installed, they'll appear in the Android app drawer as normal.
Paperize is a polished, stylish wallpaper changer that lets you switch between your pick of backdrops for the lock screen and the home screen on a schedule. You choose the pictures and the timings, and you get more control than with the default settings inside Android. The app is smart enough to be able to adjust images to fit the dimensions of your device, and there are some handy effect and filter options as well if you need them.
OpenTracks is a fitness tracking app that respects your privacy—you don't even need to connect it to the web or sign up for an account. You can use it to track runs, hikes, and cycles, and plot them on a map with details like distance covered and average speed.
What you don't get with this app is full support for wearables and fitness trackers, though you can sync data over Bluetooth from certain heart rate monitors and cycling computers. Your phone is used as the main "tracker" device, so you'll need to always have it with you.
There are no shortage of weather apps out there, but Breezy Weather makes a solid claim for being your default. It's able to pull meteorological data from numerous public sources, and then presents the data in a variety of interesting and engaging ways.
It definitely holds its own against the default weather apps you get with Pixel and Galaxy phones, and there are tons of customization options here (including which weather info gets shown and where), as well as the all-important widgets for your homescreen.
Every alarm, clock, and timer app is much like another, right? To a certain extent, but Chrono has an elegance and an intuitiveness that not many other apps of its type have. From split times in the stopwatch to time zones in the clock, there's a lot to explore.
Some of the touches I like here are the alarm filters —see active, snoozed, and disabled alarms in a tap—and the presets in the timer section, to quickly launch a meditation or workout (for example). You get full control over the color scheme of the app, too.
F-Droid does keyboards too, and with HeliBoard you have a keyboard that's fully customizable and flexible and yet which can work fully offline too—so none of your keystrokes or typing patterns are getting beamed back to a data server somewhere.
Among the tweaks you can make to the keyboard are the style and the layout of the keys, and the colors used. You can add toolbars and dictionaries, configure swipes and long presses, and add sounds and vibrations. It's comprehensive, as well as being private.
Auxio is a "simple, rational" music player that will take you back to the days before streaming music. It focuses exclusively on local files stored on your own device, which means spotty internet access is no problem, and no one is tracking your musical tastes.
All the basic features you would expect to find are here, and once you've imported your music, it's neatly sorted by its metadata into albums, artists, and genres. There's a built-in equalizer here, as well as a playlist creator, and a handful of customization controls.
Librera Reader can turn your handset into a decent e-reader—it'll be of interest to anyone who has reading to do on their phone. The app can open up a whole host of e-book and document formats, from EPUB to PDF, and has a broad range of customizations available.
You get a choice over how you read your documents and e-books—including top-to-bottom infinite scroll, left-to-right book-style scroll, and even auto-scroll—and there's all the support you would expect for bookmarks, search, and document outlines.
The driving experience with Android devices has been a bit hit-and-miss down the years, it's fair to say, with features and modes appearing and then disappearing, and Android Auto going through a few changes as well since its 2014 launch. Android Auto was originally available on both phone screens and vehicle dashboards, before being replaced by Google Assistant Driving Mode on handsets. As Gemini emerged, Driving Mode disappeared as well, which means there's now no customized interface for driving on Android devices themselves—everything gets handed off to Android Auto.
However, with the help of Headunit Reloaded Emulator (an app I found thanks to How-To Geek), you can override Google's decision and get Android Auto back on your phone. You can also beam the Android Auto interface to another Android phone or tablet. It's a $4.99 purchase, but it works as advertised—or at least it did with the Pixel and Galaxy phones I was using. If you don't have an Android Auto-compatible dashboard in your car, you can use this to get the same experience on your main handset or on a spare phone or tablet you've got.
I should note that there are some bad reviews of this app on the Play Store, though it worked perfectly fine for me. Users complain about about flaky performance and difficulties with certain phones—so it's worth reading through to see if your handset is mentioned—although others give the app a full five-star rating and seem happy with how it worked.
As always, remember the dangers of using your phone while driving. This is specifically for emulating Android Auto on a screen other than your built-in head unit—the other functions on your phone should stay unused until you're safely parked up.
Once you've installed Headunit Reloaded Emulator from the Play Store, you need to approve all the permissions that allow it to work, including Bluetooth and location access. When you're in the app, you'll see your three mode choices: Phone Mode (for using it on the current handset), and Wi-Fi and USB (for connecting to another device).
Tap Phone Mode to get Android Auto on your handset's own screen. It might not look great at first due to the cramped display size, but if you turn your phone around to its landscape orientation it should fit nicely. Any taps on the screen register just as they would on Android Auto, and you can switch apps and change the layout via the icons down the side.
To quit the interface, you need to go back inside the app rather than exiting it completely: On most Android handsets, this is now a left-to-right swipe on the left edge of the screen. A row of options shows up, including options for exiting the mode and even running it in a picture-in-picture format on your screen.
Back on the main screen, you're able to access a wealth of options under Settings. These cover everything from pixel density and screen margins to key press configurations and audio routing—you can pipe the audio from apps through your car's speakers via Bluetooth if you want, even if the Android Auto interface is on your phone.
If you have a separate phone or tablet you want to get Android Auto running on, then you need to get Headunit Reloaded Emulator up on this specific device—the one acting as the head unit (control center) in your car. You can then connect your primary phone to it, using that phone's apps and data.
The USB option is easier, if you've got a cable to spare and are happy to keep both devices hooked up during the course of your drive. Connect the cables, then on the device running Headunit Reloaded Emulator, tap the USB option on the main screen and select your main phone when it appears. On your main phone, you may need to tap the USB notification and choose Transferring files / Android Auto as the connection mode before it works.
There's a wifi option as well: You need Headunit Reloaded Emulator running on the device that's displaying Android Auto, and the companion Wi-Fi Launcher app on your primary phone (driving the maps, apps, and info on the Android Auto screen). On the former device, choose Wi-Fi as the connection method inside Headunit Reloaded Emulator, and make sure the latter device is set up to be discovered in Wi-Fi Launcher.
By default, both devices need to be on the same wifi (your car's built-in wifi for example). You can also use a direct wifi connection: select it under Wireless in Wi-Fi Launcher, and under Settings > Connectivity in Headunit Reloaded Emulator. One device will then drive Android Auto on the other—no connecting wires needed.
NotebookLM may not be the most well-known of Google's numerous AI products, but having been introduced as Project Tailwind in May 2023, it's actually been around almost as long as the Google Gemini chatbot (originally called Bard, as you might remember).
Now, perhaps in a bid to make its suite of AI tools a little less confusing, Google is renaming NotebookLM as Gemini Notebook. If you love the app in its current form, though, don't panic—not much else is changing.
"It remains a standalone product ... but it will now do more across the Google ecosystem, including inside the Gemini app and Google Search," explains Google. That "do more" will start with your notebooks appearing in AI Mode in Google Search for reference, which is rolling out soon.
Google's announcement post also mentions that users on the Pro plans ($19.99 a month) will soon be getting the upgrades rolled out last month, which were previously exclusive to Ultra plan users ($99.99 or $199.99 a month).
These upgrades include a dedicated cloud computer for each notebook you create, enabling deeper analysis of sources and the ability to write and execute code natively (via Google Antigravity). It also means more capabilities in terms of what the app can do within each of your notebooks, including creating PDFs, charts, and slideshows, and editing them through prompts.
You could be forgiven for getting confused about the difference between the standard Gemini AI chatbot app you find on the web and mobile, and the newly renamed Gemini Notebook—especially if you've not used the latter much.
They run on the same set of Gemini AI models, and can do a lot of the same tasks in terms of looking up information from the web and summarizing documents. When you've created notebooks inside Gemini Notebook, you can then also access them in the main Gemini app, confusing things even further.
What makes Gemini Notebook different is that it's very much a research tool. You load up a bunch of sources, from YouTube videos to PDFs, and then get Gemini Notebook to query and summarize those sources, with citations.
As we've written about before, you can turn your notes into audio podcasts and video overviews, as well as mind maps, flashcards, infographics and slideshows. Unlike the main Gemini app, though, it can't create images or videos out of a single prompt, without one or more sources to work from.
As an addendum: If you're already a seasoned NotebookLM/Gemini Notebook user, you may have missed the latest feature upgrade, rolled out to all users a couple of weeks ago. It's called Short Video Overview, and adds a shorter, vertical video format for your summary options: Click Video Overview then Short to use it.
There are a dizzying number of ways you can keep up with what's new on the web and try to find articles and posts of interest: Your options include the chaotic algorithms of social media, classic news portals, old-school RSS, traditional web search, AI-powered search, and whatever it is Google is trying to do with Discover these days.
It's a lot, and the latest app I've spotted to try and cut through the noise (via TechCrunch) is HyperTexting (just for iOS for now). It's a sort-of RSS reader, though it isn't marketed that way, and the core idea is that you get updates from the sites and people you want, in chronological order—with no ads, AI, or algorithm-powered interface.
As the HyperTexting app landing page puts it: "HyperTexting is your timeline minus the ads, algorithms, and AI slop. It's like the 'for you' feed, but it actually works... for YOU (not the platforms)." You can even create a HyperTexting profile of your own and contribute posts yourself, which others can then follow in turn. Here's how the app works, and why you might want to switch to its simplified way of keeping up with the web.
When you launch the HyperTexting app for the first time, you don't need to sign up for anything or open an account right away: There's already a selection of news sources included for you to get started with, which you can add to or edit as needed.
The Timeline tab is where news and updates appear, presented in chronological order. You pull down from the top to refresh the feed, just like the old days, and you're able to tap on individual articles to read them. As with any RSS reader, whether or not you get the full article text or just a snippet directing you to the website depends on the publication. Tap the toggle switch button (top right) to change the colors and other settings.
In 2026, it's actually quite strange going back to a feed like this. I felt a twinge of nostalgia slowly browsing through the headlines, without any AI-generated animal videos, adverts for mobile games that don't exist, clippings from podcast interviews, and snippets of stories I've seen a dozen times already but have just gone viral again. You get your choice of story sources, in order, whenever you're ready to browse through them. Each post comes with buttons for sharing with others or viewing it on the web, and I also like the way each source has its own profile with more ways to explore: Posts to see all the stories, for example, and Attachments to see all the visuals.
Profile pages work a lot like profiles on social media accounts, and you can follow and unfollow publications from these pages. Many sites also offer multiple feeds—including ESPN, which has different feeds for different shows and sports—so if this is available you can customize your story list further.
The Explore tab is where you need to head to find and follow new websites, and there's also a Trending tab here that shows you the 100 most-linked pages from the feeds you're currently following. If you just want the most widely appealing stories from your selected sources, this is a great place to keep returning to.
Head to Discover and you can pick out new sites you want to follow: You can either enter a few keywords to search for something, or type out a full URL, then tap Follow when the right publication appears. This seems to be the part of HyperTexting that's most obviously a 'work in progress,' as I couldn't add all the sources I wanted—including the esteemed outlet you're currently reading.
It's possible that the app is having some early growing pains, or that being able to follow a page depends on how it's hosted (the HyperTexting website says that support for WordPress sites is "coming soon"). In any case, I did find plenty of feeds I was interested in following, even if I couldn't access everything.
You can switch to the Activity tab to get back to articles you've previously looked at, like a rudimentary browsing history, while the Post tab is where you can contribute to the HyperTexting community—this feature isn't widely available yet, but eventually the goal is to have mini-blogs available for users that work a little bit like social media feeds.
There's still work to do here, but I like what HyperTexting is doing already, and it's a reminder of how discovery on the web used to be—I'll be keeping an eye on it as another way of staying updated. It's definitely worth installing if you like your new content presented in a clean, accessible way.
We're now well used to AI companies pushing out model upgrades and new features, but one of the most recent updates to Claude added something more unusual: a Reflect tool that looks back on how you've been using the AI bot.
Anthropic says Claude Reflect has been introduced to answer questions like, how often should someone use AI? How can it be used most effectively? When is AI suited to a task, and when is it better left to a human? And also, perhaps, are you letting AI do too much of your thinking for you?
In practice it works a little bit like Spotify Wrapped, rounding up your activity in the app over the past month, three months, six months, or year. And it turns out that its summaries are mostly fair and insightful, based on my testing.
Reflect is available now inside the Claude apps for Free, Pro, and Max users, though Anthropic is labeling it as a beta feature for now.
Before you can use Claude Reflect you need to have the Memory feature enabled: On the web, click your profile icon (bottom left), then Settings > Capabilities. In the mobile apps, tap the menu button (top left), then your profile icon, then Capabilities. If the Search and reference chats toggle switch is enabled, you're good to go.
When you're ready to do some reflecting, you need to be in the web or desktop apps. Click your profile icon (bottom left), then Settings, then Reflect. A summary will be automatically generated for your last month of usage, by default, but you can change the time period via the drop-down menu in the top right corner (click the circular refresh arrow icon to load a new summary).
I must admit up front that I tend to regularly delete my Claude chats as I go—I don't like leaving a long digital trail behind me for a variety of privacy and security reasons—so I can't tell you what it's like to have Claude dig through months and months of conversations. It may be that digging into bigger datasets get you different results.
What I can tell you is that Claude's review of my last month of AI use was still interesting: It correctly identified that I spend a lot of time working on little web apps and widgets that can help with my work, and of course experimenting with Claude's capabilities (also for work purposes, most of the time).
However, it does ignore your most recent conversations, depending on the calendar: If you're half way through July, your "past month" summary will only run up to the end of June, so only whole months are counted in the summary. It also seems to ignore shorter, one-off chats in favor of bigger trends.
Your reflection starts with a summary and a chart showing daily activity, but if you scroll down there's more to explore. Claude will break down your time into categories that you spend time on (like a New Tab extension for Chrome in my case), and assess your AI fluency skills too—basically, how good a prompter you are.
I scored highly in terms of "delegation" to the chatbot and "discernment" in terms of assessing the AI output, but apparently my "description" skills could use some work. Claude pointed out an occasion where I could've been more precise with a prompt, and gave me an example of how I could've refined it.
I'm not sure anything I read in Claude Reflect is going to change how I use the AI, but I think it's worth dipping into every now and again just as a self-audit. In the same way Spotify Wrapped can reveal that you're listening to far too much '90s pop, Reflect can help reassess your AI boundaries—particularly for heavy users.
I did find some of the summarizing to be a bit too polished: It's typical of AI bots like Claude to want to write a flowing narrative that can be tied off neatly, rather than a more awkward report that's more accurate, and this is evident again here. On the whole, though, it seemed to give me fair assessments.
If you do feel your Claude use has got out of hand, the Reflect summary includes a Set quiet hours and breaks link that leads to the Time and focus section of the app settings: From here you can set screen time limits for yourself with the AI, and have it remind you to take a break from prompting every so often.
The latest upgrade being pushed out to ChatGPT, heading to all users now, is GPT‑Live. OpenAI is describing it as a "new generation" of voice models for interacting with the AI chatbot, and you might find that it leads you to spend more time chatting than typing.
Voice mode for ChatGPT is nothing new, but previously it's been a relatively basic wrapper on top of the standard text input and output. It has been billed as a more natural way to engage with the AI, but GPT-Live promises to dial this fluidity up to an even higher level.
For the first time, the voice mode will be able to think in the background while continuing the conversation. It'll also give you extra space to pause when you need it, and indicate it's still listening with phrases like “mhmm” or “yeah."
You should find the upgrade on mobile and the web now (or very soon). Free users get access to GPT‑Live‑1 mini, while those on paid plans are able to access the even smarter GPT‑Live‑1 model.
OpenAI's end goal is to make talking to ChatGPT feel like talking to a real person, and GPT-Live gets closer to that. Originally, interacting with the AI via voice required a specific model for speech-to-text, another for actually responding to the query, and another for text-to-speech.
The previous voice mode in ChatGPT combined all of that into a single AI model, but it was still turn-based: You spoke, the chatbot answered, then you spoke again. With GPT-Live, ChatGPT can be talking and listening at the same time. You can interrupt it as and when needed, and responses should be faster and more nuanced.
The new voice mode is supposedly smarter when it comes to recognizing the difference between you pausing mid-thought and actually finishing your query. The model now recalculates several times a second "whether to speak, continue listening, pause, interrupt, or invoke a tool."
An added benefit of the upgrade is that even complex work and deep thinking can be passed back to ChatGPT's servers in the background, while the conversation is continuing. You can also tell ChatGPT to take a beat or slow down; visual responses have been improved as well, so you might, for example, see pop-up cards for locations, weather forecasts, and sports scores.
You can also now ask GPT-Live to translate something into a foreign language as you speak. Thanks to the new capabilities, you'll hear a running translation in the other language as you talk, with no pauses or interruptions. Improvements have also been made in terms of ignoring background noise (like background traffic or conversations happening nearby).
To get to voice mode in the mobile app, tap the soundwave-style icon to the right of the prompt box. The new mode looks a lot like the old one on the surface, but with this update, you should see Live at the top of the screen (for the time being, at least, you can tap this to switch back to the older models).
Right away, the upgraded voice mode feels more realistic and natural. ChatGPT will talk in a varied and expressive way, throwing in useful markers like "let me check" whenever it's looking something up. It'll lso hesitate and draw words out at times.
I chatted with GPT-Live for several minutes about upcoming movies, recent soccer matches, and tech news headlines, and got back answers that made sense and were respectfully brief (voice mode continues to be a refuge for those who don't want to see walls of text for every response).
There were a couple of moments where the speech glitched and the conversation hung, but that was in about half an hour of chatting (presumably these bugs will get ironed out over time). Interruptions are handled well too, with the AI pausing to acknowledge what you've said and then continuing its train of thought.
You can tweak the level of thinking ChatGPT puts into the new voice mode: Tap the sliders icon (top right), then tap Intelligence. There are three modes to pick from—Instant, Medium, and High—with varying levels of trade-off between the speed of the response and how detailed and accurate it is.
You've got all kinds of options when it comes to choosing how to communicate digitally with friends and family: not just dedicated messaging apps such as iMessage and WhatsApp, but also the DM features built into social media platforms like Instagram, Facebook, and Snapchat. Some of these apps put a lot of emphasis on security and privacy, with mentions of end-to-end encryption and disappearing messages (WhatsApp now lets you hide your phone number for example). For the ultimate in private messaging, though, there's an app you might not have heard of before.
Delta Chat is decentralized (so there's no one point of failure or control), open-source, end-to-end encrypted, and anonymous—you don't need to supply a phone number to get started. You can even message people who don't have the app. Here's what you need to know, and why you might want to make use of it.
Delta Chat has been around for about a decade now, and started as a way to keep your messages out of the clutches of big tech companies. Over the years, it's added more and more features, become easier to use, and expanded to more platforms. There have also been some changes in the way Delta Chat works as a platform.
For much of Delta Chat's existence, it's worked through email: You signed up with your email address, and the chat app was essentially a wrapper for certain conversations in your inbox. Much of the storage and infrastructure work was handled by your email provider of choice, and anyone with an email address could join in.
Now, though, Delta Chat provides an email address for you, and handles all of the necessary plumbing behind the scenes. Not only does this make the service more convenient, it means you don't need to reveal anything to use it—you won't be asked for a phone number, an email address, a name, or anything else.
That may sound like it leaves the service open to spam, but the encryption stops that: Essentially, no one can spam you without knowing your specific encryption key. Your address is a random string of characters that you can pass on to trusted contacts—there's no public directory of users, and no way of looking someone up.
There are plenty of the usual chat app features here: audio and video calls, group chats, read receipts, and the option to have your messages disappear after a certain amount of time. You can easily share files with contacts, and you can log in on multiple devices simultaneously.
You can start your Delta Chat journey through the desktop apps for Windows or macOS, or through the mobile apps for Android or iOS. As noted above, you don't need to provide any personal information to sign up for the service: Just tap or click Create new profile on the opening splash screen to begin.
The app will ask you for a username that your contacts will recognize you by, but it doesn't have to be your real name, and there's the option to add a profile picture as well. You are then taken to the main chat screen, though you won't have any chats yet—just a welcome message and a folder for your saved messages.
To add someone on Delta Chat, tap the QR code icon at the top (Android and desktop) or bottom (iOS) of the interface. Ask your prospective contact to scan this with Delta Chat on their own device, though there are also options if you're doing the scanning. You can also generate an invite link to paste into an email or another messaging app.
Everything works very much as you would expect once you're in the Delta Chat interface. Via a long press on mobile or a right-click on the desktop, you can find options for muting, pinning, and archiving specific conversations, while the new chat button (a plus icon on Android and the desktop; a pen-in-a-box icon on iOS) lets you start a new conversation or create a new group chat.
To get to the settings for an individual chat, tap the three dots (top right) on Android, tap the contact name at the top then the three dots (top right) on iOS, and click the three dots (top right) on the desktop. The options here let you set up disappearing messages, search through chats, and clear the conversation history, as well as muting and archiving.
There's a main settings screen, via the large cog icon on the desktop app interface, the three dots (top right) on Android, and the Settings tab on iOS. The options here include being able to set backgrounds for your chat, turn read receipts on and off, and configure the quality of shared media files.
There are all kinds of encrypted, secure chat apps out there, including WhatsApp and Signal—so what are the compelling reasons to pick Delta Chat instead? The fact that you don't have to offer any personal data is a big one: You don't need a phone number or an email address, and no one is going to find you (or spam you) unless you specifically decide to add them.
Then there's the decentralized aspect of it. Delta Chat's storage servers and nodes are spread out in several different locations, so one power outage or technical failure doesn't bring down the entire system. You can even host your Delta Chat data yourself, if you want.
Due to the email infrastructure underpinning Delta Chat, it's also harder to block and censor than something like Signal, and easier to get your messages exported out.
If you decide to give this app a go, there is the standard problem of trying to convince family and friends to install it too, on top of whatever they're already using—but you don't necessarily have to persuade your entire contacts list to join Delta Chat: It could work well as a private app for you and a handful of people closest to you.
If you're a regular Lifehacker reader, you'll know Proton develops a range of privacy-first products, including an email service, a VPN, and cloud storage. It also now has its own AI chatbot, called Lumo—first launched in August 2025, and recently upgraded with a host of new capabilities, including image generation.
Having given Lumo 1.0 a test run last year, I found it a capable AI assistant. While it wasn't on the same level as apps like ChatGPT and Gemini, it got close enough to make it worth considering if you didn't want to deal with OpenAI and Google. Now that Lumo 2.0 is live, I was keen to give it another go.
Lumo is available on the web, on Android, and on iOS, and you can get started for free—you don't even need a Proton account. As you might expect, there's a $12.99 per-month Pro tier available, which gets you features like a full chat history, more advanced AI models, and higher usage rates in terms of chats and file sizes. Unfortunately, the Pro tier is not included in Proton's other subscription packages—it's separate.
Read the Lumo 2.0 launch post, and you'll see the company is very keen to make its AI assistant stand out in terms of how little data and personal info it keeps. Of course, some data storage is necessary for the bot to actually work, and if you want, you can keep your chats around to refer back to.
First of all, your chats aren't used to train the AI models Lumo is built on. That's something you'll find as an option you can toggle in bots such as ChatGPT and Gemini, while it's also a selling point for Apple Intelligence. Other AIs are less transparent when it comes to this aspect, so it's good to see Proton mentioning it up front.
There's also zero-access encryption for your chats, which means no one can take a peek at what you're reading—not law enforcement, government agencies, or even Proton staff. That's a clear advantage over other AI platforms, though the likes of Google and OpenAI would argue there are strict legal procedures and strong regulations in place to keep your saved chats private. However, like other AIs using LLMs, Lumo can't offer full end-to-end encryption, as Proton explains here.
It's hard to be definitive here in comparing Lumo to other services, because a lot depends on whether you're saving or erasing chats as you go. To complicate matters further, OpenAI is currently having to hang on to a good chunk of all user chats (even the deleted ones) as part of its ongoing lawsuit with the New York Times.
What is clear is that Proton takes the issue more seriously than just about anyone else at the moment. There is no review of chats by humans, which happens with other chatbots like Gemini—in order to "help improve Google services" for you and everyone else, supposedly (which is why you should never share anything too private with an AI chatbot).
What's more, Lumo runs on open source AI models, which should mean more transparency. There's no partnership with any third party in terms of how the service is run, so you can be sure that the policies and protections Proton has implemented aren't going to be compromised through connections to other companies.
You can read more on Lumo privacy on the Proton website, including information on the large language models (LLMs) used by the chatbot. As ever with Proton, it's worth remembering that it operates out of Europe—meaning it's not subject to the same kind of data surveillance and retention policies as somewhere like the U.S.
In my Lumo 1.0 hands-on, I mentioned that image generation and deep reasoning were a couple of the features that it lacked compared to other chatbots—but that's no longer the case. To begin with, Lumo 2.0 can both generate and analyze images, so I was eager to give this new functionality a go right away.
I set Lumo to work on generating an image of a spaceship, a forest cabin, and a cartoon cityscape, and the results were pretty even: Lumo can certainly hold its own against ChatGPT in terms of the pictures it creates, and I actually preferred its space one. ChatGPT's efforts were a little bit more impactful overall, though they also went off-script (or off-prompt) more.
Lumo lets you modify images you've created through follow-up prompts like ChatGPT does, and it really impressed me in terms of how quickly the images were generated—it produced something usable in fewer seconds, though to be fair, the graphics it produced were also at slightly lower resolutions.
Lumo 2.0 comes with a new Thinking mode, which is slower than the alternative Fast mode but better at reasoning. I put it to work on a few dense scientific research papers, and its analysis and summaries appeared to be on a par with ChatGPT's in terms of how well they were written and structured. On a couple of occasions, Lumo actually presented the relevant information in a more accessible way than the OpenAI chatbot.
Web search is another area where Lumo 2.0 brings improvements, and again, Proton's AI app matched ChatGPT in terms of how it collected and organized information from the web. I got both bots to summarize a few current tech news stories, and they were equally capable in terms of summaries, complete with updates on search progress along the way, and references at the end. ChatGPT tended to be a little more thorough, but I liked the way Lumo presented its info.
Other upgrades in Lumo 2.0 include improved memory recall and management, and Custom Lumos that match the Custom GPTs in ChatGPT: They let you create siloed AI assistants for specific tasks, like polishing text or outputting web searches in specific formats. Lumo has clearly come a long way since the initial release, both in terms of the quality of its answers and the features on offer.
ChatGPT is still the more mature AI product, with features like live voice mode and professional coding assistants, but the gap between Lumo and the established AI chatbots is now significantly smaller—certainly for casual, everyday use. If you're looking for AI with a few more principles built in, it's worth considering.
Up to this point, adding someone on WhatsApp has meant giving them your phone number—which is perhaps not something you always want to divulge, especially if you only just met, or you're in a group chat with lots of other people.
That's finally changing: WhatsApp is embracing usernames. The update is launching "later this year," according to WhatsApp, though you can start reserving your name of choice this week. It'll work much like usernames do on other messaging and social media platforms, meaning you'll no longer need to give up your digits to chat with someone.
"Usernames are our latest step to make WhatsApp even more private," explains a recent WhatsApp blog. "There's no directory to browse and no suggestions—people will need to know your exact username to contact you for the first time."
There's also an extra level of security just in case you think someone else might stumble across or guess your username: There's an optional username key you can add, which other people will need in order to be able to add you.
WhatsApp didn't offer precise timings on the rollout, but it says usernames will start appearing "gradually over the coming months" and that users will get alerts when the feature is available in their region. Once usernames are enabled, they'll replace phone numbers when making new connections.
Creators, small businesses, and organizations can claim the same username they have on Facebook or Instagram. For everyone else, you can head to Settings in the mobile app and choose Account > Username—though the option doesn't seem to be live yet for everyone (if you don't see it, try updating the app).
And just in case you were thinking about reserving @TomHolland or @Zendaya, WhatsApp told The Verge that celebrity and politician names have already been reserved. If you're completely stuck for inspiration, there's going to be a username generator you can make use of, too.
The announcement blog post specifically says reservations will be open "starting this week," so if you're keen to bag your username of choice before anyone else, keep a close eye on WhatsApp in the coming days.
NotebookLM is Google's lesser-known, research-focused companion to Gemini, and it just got a hefty upgrade that makes it more useful than ever. If you need to go deep on research, study, reporting, or analysis, it's one of the best tools available right now.
Here's what's new in the latest update: NotebookLM can now write code (via Google's Antigravity development platform), and output in a variety of different file formats including PDFs, PNG/SVG charts, Excel spreadsheets, and PowerPoint presentations. You can make edits to these files after they're created, too, with follow-up prompts.
If you don't have any source material to begin with, you can chat this through with NotebookLM and it'll work with you to find some suitable sources on the web—right in the conversation you're having. It's a more natural way of approaching a topic, if you're starting from scratch.
The AI will show you its reasoning as it goes, so you get regular updates on what the AI is doing or thinking about, rather than just icons showing spinning dots. In fact, it's a lot more like the standard Gemini app, while still maintaining that focus on research and study.
These upgrades are rolling out now if you're on a Google AI Ultra plan (that's the $100 or $200 per month one), and Google says that they plan to expand to "others over time" without being any more specific than that. I was able to test out the new features using a Google AI Ultra plan.
The traditional way to get started with NotebookLM is to load it up with sources—PDFs, web links, YouTube videos, and the like—and then start asking questions about the information you've amassed. Now, though, you can dive right into a conversation with the AI bot and pick up some sources on the way.
To put this to the test, I wanted to see what NotebookLM could find online about the films of Christopher Nolan. Not only did the AI tool find lots of relevant info, it also synthesized it into a helpful overview. At the end of the overview, I got an Import button, so the sources that NotebookLM was referring to could be added right away. What's more, the initial reply gets added as a source, too.
Previously, NotebookLM was capable of searching the web for sources on your behalf, but the process was split between a couple of different panes in the interface, and required several clicks to get everything imported. Now, it's integrated much more neatly into the main chat box, so you can get researching and analyzing faster.
You still have the ability to keep adding sources in the manual way via the left-hand pane, and as always NotebookLM will refer to these sources in its answers. If you ever wonder where it's got a particular idea or fact from, you can click the citation button to see. Here, a lot of the heavy lifting was done by Wikipedia.
While I couldn't see any hallucinations here, there was a weird quirk where a request for the "key" cast members of Nolan films returned a lot of men but just two women from his 12 movies: Marion Cotillard and Anne Hathaway. Apologies to Carrie-Anne Moss, Hilary Swank, Katie Holmes, Scarlett Johansson, Rebecca Hall, Elizabeth Debicki, and the (Oscar-nominated) Emily Blunt.
I think it's because NotebookLM became fixated on a list of frequent Nolan collaborators on Wikipedia, but that doesn't fully explain it. Something to bear in mind if you're relying on AI for your research: Even if actual mistakes are rare, the summarizing and collating that these tools do still needs checking.
With a few more prompts to increase the female representation in this Nolan movie project, I wanted to test out another feature: the ability to create editable files. You can request Microsoft Office files, PDFs, and charts in PNG or SVG format, and get a download you can open up.
I asked for a PowerPoint slideshow covering all of Nolan's films and NotebookLM obliged: I got a well-formatted, typo-free introduction to these movies, including the premise and themes of each one, and the key cast members. What NotebookLM still can't do is generate images or pull them from the web, so you'll need to do this yourself.
Something else you can now do in NotebookLM is request edits to files that have been created, Nano Banana-style. For my PowerPoint, I requested changes to fonts and background colors, and NotebookLM was able to follow my instructions exactly—giving me a shareable slideshow from 10-15 minutes of work that would otherwise have taken me at least a couple of hours.
Next, I tried creating a PDF primer on Nolan movies, and while its design certainly isn't going to win any awards, it was reasonably well laid out in a clean and consistent way. The information was accurately pulled from Wikipedia and the British Film Institute, and again, it's a real time-saver—the days of having to copy, paste, and reformat answers from NotebookLM look to be gone for good for a lot of tasks.
Switching to landscape from portrait PDF format was well-handled by NotebookLM, and while minor edits are trickier without a visual interface—moving a line, adding a new text box—you can always do these edits manually as normal if you need to. Or, if you're editing an Office file and want to go all-in on AI, Copilot will be happy to help.
What I didn't try here are any of the new Antigravity coding features, as that's really a whole different topic, and something I don't have much experience with. In terms of the rest of the new features here, NotebookLM is clearly better and more useful than ever, even if the perennial AI warning to double-check responses still applies.
Android 17 is on its way to Pixel phones around the world, and while there aren't a huge number of major upgrades—Google now usually pushes these out through the year—there is a useful new feature that gives you easier access to your apps. The feature is called Bubbles, and it's an extension of the chat bubbles that rolled out with Android 11 and that have since largely been forgotten about. This time the bubbles are coming to all categories of app, so you can set a floating shortcut to appear at the side of the screen containing the apps you use the most.
This shortcut expands and collapses as you tap on it, and you can have up to five different bubbles on the go at any one time. You can put your bubbles anywhere you want on screen if you're using a regular phone, but if you're using a foldable they get docked to the bottom right corner. It's like having an extra mini launcher to call on.
The clever part is that these bubbles aren't simply another way of launching the app: They actually give you access to it in a slightly smaller window, so you can quickly check up on something or interact with an app before going back to whatever you were doing before. So if you've writing a text message and you've got a website you need to refer to in a bubble, you can quickly get an overlay of that site, check what you need to, and then get back to the message.
It promises to make multitasking much more intuitive and smooth on Android, and I've been using it extensively since I got the Android 17 update. I found it did take me a little while to figure out how I could best make use of it, but it's already changing the way that I use my phone.
You can create a bubble for an app by long-pressing on its icon on the home screen or in the app drawer, and choosing Bubble. If the app has its own long-press menu, then you'll just see the bubble icon without the text—it looks like a rectangle with an arrow pointing to one corner of it.
Tap on your new bubble to open the full selection, and you'll see there's a + (plus) icon for adding more bubbles as well. Tap and hold, and you can drag the bubbles shortcut to a different part of the screen. To get rid of a bubble, tap and hold on it, and drag it down to the X at the bottom. You can do this for all your bubbles at once, if you need to.
There are some apps that I can't create bubbles for, and I'm not sure why. It's possible the developer hasn't enabled the feature, or perhaps they're just incompatible with bubbles. They included the Google Authenticator app, the games I've got on my phone, and the default Camera app for Pixels.
When you select a bubble, you get the app overlaid on top of whatever else is on your screen, and you can interact with it normally—it's just that the window is slightly smaller than it otherwise would be. You can then tap outside the window to close your app bubbles, or use the row at the top to switch to a different bubble.
This doesn't replace launching an app in the normal way—you can still do that with apps that you've made into bubbles—it just gives you an alternative, quicker way of accessing certain apps without losing what you were already doing. It's a bit like a more agile recent apps screen, or Slide Over on the iPad.
I've got WhatsApp, Pocket Casts, and Google Health set up as bubbles, so I can quickly check and reply to messages, start and stop podcasts, and look up my daily steps without too much of an interruption. While doing these tasks and going back to a previous screen would usually take several swipes and taps, now they just take a couple of taps.
Google's efforts to transform online search and the ways we find information on the web continue, and there's now a new feature available in AI Mode: information agents. This new feature will keep tabs on news for you and ping you when needed. (At least, that's the idea.)
The update was announced at Google I/O 2026 back in May, but it's now appearing for those with a Google AI Ultra subscription (that's the $99.99 or $199.99 per month one). Google's Robby Stein says that more people will be getting access in the summer.
Google hasn't been any more specific than that, but presumably this will filter down to the other paid plans in the next few months, and eventually to everyone. I'm making use of my Ultra subscription here to get these information agents running.
This Tweet is currently unavailable. It might be loading or has been removed.
Here's the idea: You look up, say, news on the next James Bond movie. Then AI Mode can keep tabs on the web and ping you whenever there's news on a casting decision, a release date, or a trailer. It saves you having to run multiple searches because the information comes to you (it's a bit like the old Google Alerts, if you remember those). These updates appear both in the AI Mode section of Google search and in the Google app on your phone. Or at least they're supposed to.
To set up your information agents, you can head to Google search on the web, then switch to AI Mode via the button in the search box. You then launch your search like you would if you were chatting to Gemini—something like "tell me who the main stars are in the upcoming Christopher Nolan film The Odyssey."
After you've got your answer, which is hopefully hallucination-free, you might get asked at the end if you want to set up an information agent to keep you updated on whatever it is you've searched for. If you don't get asked, you can put in the request anyway: Tell Google to "keep you updated" or words along those lines.
You'll receive a confirmation message, and then whenever new information appears online, you'll get pinged about it. Updates appear as notifications in the Google app on mobile, and as new entries in the original AI Mode conversation—so if you delete the chat, you stop getting updates (you can access your previous chats via the AI Mode history button on the left.
Google doesn't say how often you'll get updates, but in my experience it was... not at all. I set up an information agent to keep me updated on the latest World Cup scores and group standings, which I thought was a fairly straightforward task, but my Google app stayed stubbornly silent and the AI Mode chats stayed static over a whole day as the goals went in and the matches went by.
Looking around the web, it seems that AI Mode information agents are working for other people, so this appears to be an isolated bug that may well get ironed out in a few days—but it's frustrating to have these features launch and then not work properly. I've had the same experience with Gemini Omni too, so maybe it's just me.
What did work was telling the information agents to give me a daily summary at a specific time. When I did this, I did indeed get a delivery of the latest World Cup news and everything that had happened in the tournament over the past 24 hours, both in the Google app and as an update to my conversation with AI Mode.
In fact, this might be a better way to use this rather than expecting updates at random times throughout the day. I can definitely see myself using it for big news topics I'm interested in, and to pick up stories I might otherwise miss, though it doesn't seem to be set up for important breaking news.
Titles like Microsoft Flight Simulator can offer truly gorgeous visual experiences, but there's an awful lot of simulation involved before you get to the flight. You have to learn a lot about cockpit controls, take-offs and landings, and other simulator-y stuff before you're allowed to soar into the sky and see the planet from a bird's eye view.
That's one of the reasons that the flight simulator mode that just launched in Google Earth on the web is likely to be so popular—in just a couple of clicks, you can be zooming through canyons or across cities, without worrying about what kind of plane you're in or how your flaps are set.
As Google mentions, this isn't a completely new experience. It's actually been ported over from the Google Earth desktop app, but there, it's rather hidden—more of an Easter egg than a full-blown feature—and offers much more of a flight simulator feel. For example, you need to choose a plane, and figure out how to take off in it, before you get to the eye-popping visuals.
This Tweet is currently unavailable. It might be loading or has been removed.
Now that this simplified version is available in the web app for free, and it can be run instantly without any flight training, it's likely to be much more popular. To test it out yourself, point your browser towards Google Earth on the web, then click Explore Earth. Find somewhere you want to fly through, then choose Tools > Flight simulator.
Google has helpfully provided some instructions for flying around. You can use either the mouse or the keyboard, and there are basically four directions to master: Up, down, left, and right. You can also adjust your thrust speed.
Everything works through your web browser, so you'll need a fairly modern Windows or macOS computer for this to work smoothly—a lot of data needs to be streamed to your system as you fly. "Flying at extreme speeds or over low-bandwidth connections may result in temporary loading delays," Google says.
The first thing you'll notice once you're in the air is that your flight is harder to control than you might think. It's worth taking a few seconds just to appreciate the view before you do anything, because there's a fair chance you're going to get into trouble pretty quickly once you start pressing keys or clicking around.
Use the Up arrow to nudge the nose of the plane down (go lower), the Down arrow to nudge it up (go higher), and the Left arrow and the Right arrow to bank in those directions. Small touches are key: If nothing seems to happen immediately, don't keep pressing the keys, as you'll very soon get out of control.
I found the mouse controls, where you just click and drag on screen, even harder to use. Within a few seconds of trying to change course, I was flying upside down. Not long after that, I was spinning around uncontrollably. Every adjustment I tried to make seemed to make the situation worse.
Don't worry about too much about hitting the ground—you'll just see a "You crashed! Restart" message, and you immediately take off again. Actually, it was quite a relief after a few minutes of spin-induced dizziness. I think Google Earth needs to make the flying controls less sensitive, or at least give us a quick reset option.
For now, you can't pause or reset the flight without quitting the simulator (via the arrow, top left) and going back in. There are no external views available either—you're basically locked in to looking dead ahead from the cockpit
Even with those caveats, and especially now that I am starting to get the hang of the controls, it's fantastic to be able to soar across any point of the globe in 3D. If you've been looking for something like Microsoft Flight Simulator without the learning curve—and you don't care about advanced features beyond the actual flight—this is a perfect way to zone out for a few minutes.
Among all the other artificial intelligence upgrades Apple is rolling out for us this year, you'll find that there's a significant jump forward in Image Playground's AI image generation abilities. Before now, the app's outputs were rather limited in terms of size, style, and possible prompts. It was very much AI images for beginners, with the results mostly basic and generic.
With iOS 27, iPadOS 27, and macOS 27, that's changing. You can try the developer betas now (though you probably shouldn't; the public betas are arriving in July), and Apple says the full releases will appear later in the year (most likely around September). Image generation is something both ChatGPT and Gemini have been steadily improving, to the point where some generations are difficult to distinguish from real photos. So how does Apple's new and improved Image Playground compare?
Open up Image Playground on the newest versions of iOS, iPadOS, or macOS, and you'll see there are several new capabilities. First, you can ask for photorealistic images in the prompt box, as well as the sketches and illustrations previously possible: Ask for a photo of an English meadow or a towering temple, and Image Playground will oblige.
There's more control over your images, too. You can submit a photo as part of a starting prompt, and you can choose between square, portrait, and landscape orientations for your picture—these options weren't available before. If you want to transport your pet dog to the jungle, that's possible now.
Then there are clever editing tools, similar to those in Google's Nano Banana AI model, that let you change specific parts of a generated image without having to render it entirely again. You can use a follow-up prompt to request changes, and even highlight the part of the picture you want to edit.
You can now change the color of objects, remove objects altogether, change the weather of a scene, whatever you can think of, really. Based on my testing, it all works well, and produces results that look impressive and consistent. I managed to put a cartoon-style cat in a street, and then change the cat's color without affecting the background.
Apple says that everything created through Image Playground will have the same SynthID watermark as generations from Google Gemini and ChatGPT, and we know because of Private Cloud Compute that no user images will be stored or accessed by Apple, or used to train any of its models.
Finally, you can do more with these images too—setting them as Contact Posters or wallpaper for the lock screen, for example. You can find more AI image tools in the Photos app, where you can apply edits similar to those possible in Image Playground to whatever's in your photo library.
Image Playground is much better—no doubt thanks to a boost from Gemini—but even still, the images aren't quite up to the very high bar that Gemini and ChatGPT have set now. You can see below how my request for "a photorealistic image of a small, ancient-looking spaceship floating between the stars, with an Earth-like planet behind it" was interpreted by Image Playground (left), Gemini (center), and ChatGPT (right).
Apple does okay, but to my eye, Gemini and ChatGPT generated results that are more immersive and detailed—like something you'd see in an actual science fiction film. There's more detail and more imagination, although to Apple's credit, Image Playground rendered the fastest.
For the next challenge, I tried asking these AI tools to "move my cuddly toy from my floor to a pebbly beach, with the tide lapping at its edges." Again, all these attempts are good, but Gemini (center) and ChatGPT (right) add extra layers of verisimilitude in terms of color, angle, and texture (though Gemini seems to have created two shorelines).
I asked all three models to remove the toy and just leave the beach, and they all managed it more or less perfectly. These are Photoshop-level edits that used to take me hours, but can now be completed in seconds. It's truly impressive. Gemini and ChatGPT still have the edge in terms of quality, but Image Playground comes built into billions of Apple devices. It's now good enough that many users will likely choose not to switch to something else when they need to generate an image with AI, which might make all the difference.
Apple spent a lot of time talking about the upgraded Apple Intelligence platform and the new Siri AI app at WWDC 2026, and in the days since, a few more details have emerged about how the AI model partnership between Apple and Google will affect the new software—but answering the question of how much of Siri AI is Apple, and how much is Google, is still complicated.
Back in January, we got official news that Apple would be tapping into Google's Gemini AI models to help power Apple Intelligence, that the deal would last multiple years, and that Apple's "industry-leading privacy standards" would be maintained.
Neither Apple nor Google explained much at the time about how this partnership would actually play out, but it was clear that this was more significant than Apple's earlier ChatGPT deal, where Siri simply shunted off prompts it couldn't reply to.
I expect there was plenty of debate within Apple about whether a technological deal with an arch-rival was worth it, even if it meant catching up more quickly with its AI. Ultimately, CEO Tim Cook and his fellow executives decided that it was—and after WWDC 2026, we have more information on the details.
Over to the WWDC 2026 keynote, where Apple's senior vice president (SVP) of software engineering Craig Federighi told us the vastly improved Apple Foundational Models (AFM) had been developed through a "deep collaboration" with Google. Apple had been "leveraging" the technology behind the Gemini models, in Federighi's words, to create the AI that now powers Siri AI and the other new Apple Intelligence features.
And you can certainly see the Gemini influence: Apple's AI is now truly multi-modal, capable of processing audio, voice, and text, and much better at producing text and images of its own. Image editing is much improved—very similar to Nano Banana 2, you might say—and Apple's AI now has much better world knowledge too, which is another area where the Gemini models excel.
However, these are still ultimately Apple's own models. For local models, we have the on-device AFM 3 Core and AFM 3 Core Advanced models, which sit on iPhones, iPads, and Macs—though the latest hardware is needed for for the Advanced version (the one that allows you to tweak Siri's pace and expressiveness). Per Apple, AFM Core Advanced needs an iPhone Air, iPhone 17 Pro or Pro Max, an M4 iPad or later with at least 12GB of RAM, or an M3 Mac or later with 12GB of RAM.
In follow-up comments (via 9to5Mac), Federighi said, "we don't have the Gemini app as our app." That is, Apple Intelligence doesn't use the Gemini AI models, or the client code Google Gemini app users get, or a knowledge base built from Google Search. All that work has been done by Apple.
There's no doubt Apple needed the Gemini AI models to get its own models up to par this quickly, but Apple executives are understandably keen to not make too much of the partnership, for the same reasons that they won't talk about the billions of dollars Google pays each year to remain the default search option in Safari.
Federighi still had time for some barbed comments though: "Some appear to be racing forward, seemingly pursuing AI for the sake of AI, without clear regard for the people—all of us—that it's ultimately meant to serve," he said, in a dig at the competitors that have left his company in the dust on AI (while Apple has paid out $250 million in settlements for promising AI features that never showed up).
That covers the local Apple Foundation Models, which keep all of your queries and data private and protected on whatever device you're using. It's with the cloud-based models that the waters get a bit muddier—these are the models that are called in to deal with larger, more complex tasks that can't be handled solely on-device.
As described by the Apple team (via Ars Technica), the AFM 3 Cloud is for general-purpose use. Then there's ADM 3 Cloud for image generation, and AFM 3 Cloud Pro for "more sophisticated" queries (and, it sounds like, the beginnings of agentic work). Like the on-device models, they use some Gemini smarts at the most basic levels, but with Apple's own contributions and tweaks on top.
Those first two models run on Apple servers, but queries sent to AFM 3 Cloud Pro are going somewhere else: They'll be sent to Google data centers, to be processed by Nvidia GPUs. However, according to Apple, the exact same Private Cloud Compute (PCC) protections will be in place for those data centers as for those run by Apple.
That means no data is stored (it'll be wiped after the query is processed), no one else can see it (not even Apple or Google), and your identity is masked. Apple lets third-party security auditors check its PCC code, and that's going to be the case with the AFM 3 Cloud Pro models and Google's servers, too.
There is one small wrinkle: Apple says "PCC on Google Cloud will be gradually ramping towards the complete set of protections throughout the summer preview period." So if you're running one of the developer betas, some of your most complex AI queries might not yet be as fully protected as you would like them to be.
Apple is promising more technical detail on all of this as we get further towards the full launch of the new software updates and Siri AI, but based on the information we have now, it seems to have managed a balance between boosting its AI with Gemini while retaining all of the Apple-ness that its users are going to expect.
There was a lot to take in from Apple's WWDC 2026 keynote, but as is the case every year, the presentation couldn't hope to cover all the changes, new features, and upgrades to Apple's software. Developers are already busy installing the first beta versions of Apple's new operating systems and seeing what they're capable of—so, in the interest of trying out the new features of macOS 27 as early as possible, I've dusted off my developer account to give the new "Golden Gate" update a try.
An Apple developer account is required if you want to access the beta right now (as is the case with iOS 27). Getting a developer account, which allows you to install early versions of Apple's software before it is released to the general public, used to require you to pay $99 per year, but that's no longer the case—you can receive early access to the developer beta for free by registering your Apple Account with the program. Installing these early betas is not recommended for most people, as you're likely to come across numerous bugs and glitches, some of which could have serious consequences for the apps and data on your devices. The most cautious strategy is to wait for the full release this fall, though a more advanced public beta is also due in July.
Personally, I haven't experienced any issues of note in my time with Golden Gate, and even at this early stage, I can see the promise of Apple's latest desktop software update. In particular, there are five features in this update I'm glad to see, and that you should be looking forward to arriving on your Apple Silicon-based Mac later this year.
As someone who always has plenty of tabs open, I'm intrigued by the new automatic tab organization tool that's available in Safari for macOS 27. It might even make me spend more time in Safari compared to Chrome.
"Safari analyzes each page, identifies similarities, and then brings related tabs together," Apple says, so you might get all your puzzle web apps in one group and all your basketball news articles in another. You can set up the feature by clicking Safari > Settings > Tabs and then choosing By recommended topics or Automatically for Organize tabs.
I've found the feature to be of limited use so far (this is a beta after all), but there's potential here. At the moment, Safari does a good job of automatically grouping together tabs open from the same website, but doesn't reliably recognize tab topics or group them together that way. I expect this will improve with time, as Apple made a big show of explaining it in the keynote.
This probably has niche appeal, but personally speaking, it's something I'll be making plenty of use of: While macOS previously supported ultrawide monitors, the new Golden Gate update will offer broader support for a greater number of native resolutions and refresh rates, with sharper scaling included.
This means if your ultrawide monitor has always been a bit glitchy and limited when connected to a MacBook (like mine), you should get a far better experience this time around. What's more, macOS 27 will remember layouts on ultrawide monitors as well, even after disconnecting and reconnecting, so no more scattered icons and windows.
iPhone mirroring is already one of the slickest and most handy features on Macs, and it gets better with macOS 27. On supported apps, you can drag out the iPhone window to a dual-pane, iPad-style layout (no doubt in preparation for the upcoming foldable iPhone).
Another neat addition is that you can now access the iOS Control Center through iPhone mirroring, so if you need to turn on airplane mode, start a screen recording, or do anything else through Control Center, it only takes a couple of clicks (View > Control Center).
A year after Liquid Glass was thrust upon us, there's finally a proper system-wide customization option: A slider that lets you adjust how much Liquid Glass you want in macOS. I've never had a huge problem with Liquid Glass, but I'm still happy to have the option to adjust it to suit my own preferences.
You can find the slider under Appearance in System Settings, and there's a neat little preview thumbnail above that shows you how the adjustments will look. Right now, the difference from one side of the slider to the other isn't huge: It goes from "fairly translucent" to "hardly translucent at all."
Based on the demos we saw during WWDC 2026, Apple now has an AI chatbot ready to offer Gemini and ChatGPT some genuine competition (with a little help from Google's AI models). Sadly, at the time of writing, I'm still on the waiting list for the new and improved Siri, so I can't tell you what it's like to use.
Still, I'm intrigued by how Apple has integrated Siri AI into the desktop OS. Whenever you open a Spotlight search, Siri will pop up to answer your questions, analyze something on screen, browse through your photos, send emails, pull information from the web, and more. As on iOS, it now comes with its own dedicated app as well, where you'll be able to sort through your chat history to review previous conversations with the chatbot.
Even if it can't quite match the capabilities of its rivals, the fact that it comes built into Apple hardware—and will sync chats across devices too, for the first time—gives it a big advantage in terms of convenience.
Google Keep is one of the best note-taking apps around, managing to cram a lot of useful functionality into a clean, simple interface that's basically a collection of colored digital sticky notes you can put to whatever use you like. It couldn't be easier to create or access notes on Android, iOS, or the web, but even if you're a long-time Google Keep user, you might not have found all the features available just under the surface—and this list of tips and tricks will get you caught up.
This one tops the list of unknown Google Keep features that most anyone can appreciate: You can use it to share notes with other people, just like you're sharing a Google Doc. There are lots of ways this can be helpful, including family note-taking—for vacation packing, for example, or managing grocery lists.
If you've got a shared grocery list for the family, then there's less chance of someone buying milk when someone else has already picked it up, for example. To share a note, click the collaborator button that looks like a portrait picture (on the web), or tap the three dots (lower right) then Collaborator (on mobile).
This is a handy feature that's unfortunately rather limited: It only works on Android, only applies to list notes, and requires an AI subscription plan from Google. But if you meet all of those criteria, you can get Gemini's help with lists: Tap the + (plus) button, then List, then Help me create a list.
You can then prompt Gemini to create a list for you—things to do on a trip to New York, or a back-to-school shopping list for a 10-year-old, or the best horror movies of the 1970s, or a to-do list for planning a 30th wedding anniversary. The prompt can be just about anything you like, and Gemini will then return a list for you to work with.
You can quickly create drawing notes in Google Keep on the web and mobile, but what you might not have realized is that if you write notes using your finger or a stylus, the app can read the text within them. The handwritten text will be included in Google Keep search results, provided your writing is legible enough.
Version history is something you don't always get with this category of app—looking at you, Apple Notes—and in Google Keep, it means you can go back to earlier versions of a note if you need to. Maybe you've checked off a whole number of items you shouldn't have, or want to revert to the earlier version of an idea, or you want to reuse the entire list.
To use this feature, you have to be using Google Keep on the web. Click the three dots by a note, then Version history: You get a list of previous versions of the note, and can download them in HTML form. There's no way to actually overwrite the current note with an older version, but you can copy and paste the text in from the HTML file.
This was one I didn't find for a long time because I tend to mainly use Google Keep on the web, and this option is exclusive to the mobile apps. If you tap the + (plus) button then Audio on Android, or the mic icon at the bottom of the interface on iOS, you can create an audio note (which then appears on the web app too).
This is best for recording quick thoughts when you're on the move and can't type out a note like you normally would. On the newly created note, you get a mini player to listen to the audio, as well as a text transcription of what you've said (which you can edit if something wasn't accurately transcribed).
Google Keep has a labels system that lets you keep your notes in the order you prefer: You can add or edit the labels associated with a note via the three dots in the corner, and the labels you use are entirely up to you. You can then navigate by label via the left-hand navigation pane (click or tap the three horizontal lines, top left, if you can't see it).
But there's hidden functionality here: Each note can have multiple labels, which gives you even more flexibility when it comes to configuring your filing system. You could have labels for both "work" and "personal" notes, and then subdivide those categories with further labels—so maybe "vacation," "shopping," "money," and "family" in the personal category.
Select the search box at the top of the Google Keep interface and you'll see a Things section (click or tap More to see the full list). This vague header collects labels that the app has applied in the background based on the contents of your notes, and can include categories such as Groceries, Places, Travel, and Movies.
Even if you don't have your own label system in place, you can use these automatically applied categories to sift through your notes—maybe if there's something you've written down but can't remember where. For my notes at least, the searches seem to match up pretty well with the content of my notes.
If a note starts to get too big and complex for Google Keep, you can turn it into a Google Doc without affecting the original note. On the web, click the three dots in the corner of a note, then choose Copy to Google Docs. You can then edit the new document in Google Drive independently of what's happening in Google Keep.
If you're a heavy Google Keep user, notes can quickly pile up and get lost. One way to make sure you don't forget about a particular note is to flag it with a reminder. Tap the bell-shaped icon along the side of a note on the web or on mobile to specify the time and date of the reminder, which can be set to recur if needed.
You'll see the reminder information on the note itself, but the actual reminding is handed off to Google Tasks—if you open up Google Tasks, you'll see a reminder has been set. You can edit and delete reminders from Google Tasks as well as from Google Keep, with the changes synced across as needed.
Google Keep doesn't show up on the Android Auto dashboard, but that doesn't mean you can't take notes in the car—you just need to use Gemini instead. First, open Gemini on Android, tap the menu button (top left), then the cog icon, then Connected Apps to make sure Google Workspace (including Google Keep) is linked.
While in your vehicle, you can then say "hey Google" and issue a command like "create a note that says remember to buy flowers" to add a note to Google Keep. You can tap into the generative AI powers of Gemini, too—for example, you could ask it to create a checklist of ingredients for a lasagna while you're on the way to the grocery store.
Much of my day is spent sitting at a computer writing, and in recent years I've most often accompanied this by long (or live) YouTube videos that offer background sound without being too distracting. There are all kinds of options: scenic railway journeys, TV show tunes, piano instrumentals of songs I like, the sounds of forest rain, movie soundtracks, walks across game worlds, and more.
In recent months though, AI-generated mixes with AI-generated thumbnails have become much more prevalent. Run a quick search for study and chill-out music and you'll find plenty of videos where the artwork looks suspiciously like something ChatGPT would make and the audio track is what you'd expect to get out of an app such as Suno.
It's harder than ever to spot AI-made content, especially when it comes to simpler, more minimal creations—like illustration-style images or lo-fi chill-out music. I don't want to listen to AI music, so at the start of each day I'm now clicking around warily on YouTube trying to find something that has been composed and packaged by actual people. It's not easy anymore, but it's still possible.
I'm not completely against the idea of AI, though I think there are some major problems with it that we're not properly reckoning with. Gemini AI might give you a better search result for "the best restaurants in San Francisco for young kids" than a list of 10 blue links, but it still relies on human experience and writing. The AI has never had kids or been to San Francisco, so what happens to those results when actual people stop writing and publishing on the internet?
When it comes to music, I don't want to listen to tunes put together by machines, based on algorithms and the mashing together of real work done by real artists. You could argue that it doesn't really matter so much for background electronica that's being put on while working or studying, but the principle is the same.
There's a line in the Westworld TV show where one humanoid robot, virtually indistinguishable from a real person, asks the question: "If you can't tell, does it matter?" We're now at the stage where we often can't tell the difference between AI and human content, but I'd argue that the difference does still matter—and matters a lot.
Aside from all the considerations about energy use, environmental damage, and copyright infringement that come along with AI (and which would all take an entire article to cover), I think there are numerous ways that the tech can be helpful. When it comes to art and music, however, I want my clicks and listening time to support actual artists.
It's something that YouTube is aware of. On some videos you'll now see a How this content was made section, disclosing the use of AI. The problem is, this relies on either the content creator owning up to it, YouTube's own AI tools being used, or AI watermarks being included in the files. Based on what I see on the platform, I don't think much of the AI content is being flagged.
So I'm left in a situation where I'd rather not listen to channels where the artwork or the music is AI-generated, but it's difficult to spot something that's made by AI. What I've started doing is looking for the channels that are definitely curated and produced by human beings, rather than trying to identify subtle signs of AI.
You'll actually see it a lot in channel titles and video descriptions now, so you could just search for "no AI" or "AI free." It's also worth digging into the descriptions to look for links to the actual music used and the artists who are being supported. Check up on the history of the channel, too—what other videos does it offer? How are they made? If there are real flesh-and-blood human performers in the video, then that's ideal.
As already mentioned, YouTube has its AI labels, but I wouldn't rely on them to any great extent. If you can't find evidence of how the music is being made or who's behind it, and there are no links to actual recordings or artists (or footage of the music being made), then at this point I think it's safer to assume it is AI rather than not.
One of the best and longest-running channels in the business in this category is Lofi Girl, which has been around since 2017, way before the generative AI boom. It was founded by a real record producer (Dimitri Somoguy), with an iconic character drawn by a real person (Juan Pablo Machado)—you can read about it on Wikipedia.
There are also now a growing number of channels that position themselves as containing no AI. One of my favorites is Yellow Cherry Jam: The videos here feature a man, a woman, a dog, and plenty of scenic backdrops. It's all very relaxing—and real.
I also like Coulou's Vinyl Cafe, where our man Coulou wanders around his apartment putting on one great record after another. The music is all listed and shown off in the video, and as good as AI video generation has become now, there's no way it could create an hour of this without a chair leg disappearing or a jumper changing color.
Judging by many of the comments under those videos I've linked, AI-free music and AI-free YouTube channels are something a lot of other people are looking out for too: They are there, if you look for them. I've now built up a long enough playlist that I'm confident of not running into AI anytime soon, and it sounds great.
Apple has struggled to keep up with the likes of OpenAI, Anthropic, and Google with its AI technology, with the long-promised Siri upgrade repeatedly delayed. It's now trying to fix that: At WWDC 2026 today, we got details of the biggest Apple Intelligence upgrade since it launched, with a focus on understanding users and keeping data private.
The updates will roll out across iPhones, iPads, and Macs, and was announced by Apple's senior vice president of software engineering Craig Federighi—who started with a dig about some AI companies moving too fast and neglecting the priorities of users (not a bad way to reframe Apple's previous problems with AI development).
As we knew already, Apple has partnered with Google to develop its new Apple Foundation AI Models, though you won't see a Google logo or "powered by Gemini" message anywhere. Apple has tailored the Gemini models for its own purposes and says there's now improved understanding across text, voice, and images.
There are four central components to these new models: Personal context (knowing about you and your preferences), world knowledge (drawing in data from the web and other sources), actions (being able to complete tasks inside apps), and on-screen awareness (knowing what you're doing on your device).
Some of the examples Apple showed for personal context included knowing about upcoming birthdays, favorite recipes, and locations you've visited. All of this information could be pulled into AI responses or routines to make them smarter.
Different models will run on devices and in the cloud, and be kept fully private—so not even Apple will be able to see your data and AI interactions. As has always been the case since Apple Intelligence launched in 2024, Apple uses what it calls Private Cloud Compute to only process data in the cloud when it's absolutely necessary, and to delete the data right after the processing is done.
These privacy protections are audited by third-party security specialists, Apple says, so you can be sure your data isn't being analyzed or kept. The company is clearly going to continue to lean on privacy as a way of differentiating itself from its competitors.
In terms of the specific benefits you'll see from these new AI models, you should get smarter, faster responses in apps and in the newly upgraded Siri app. These technologies are going to power Apple AI across all of its products—enhanced by a new System Orchestrator that was flagged up by Federighi—and will work everywhere from enhancing photos to dictating text.
Of course, this isn't the first time Apple has hyped up future upgrades coming to its Apple Intelligence technology—we won't be able to assess these changes fully until they're pushed out into iOS, iPadOS, and macOS.
Letterboxd has created a cozy online space for film lovers to get together—those who truly appreciate the art form, and who want to compile lists, aggregate ratings, and share their movie opinions with others. Its success has led to other similar platforms appearing in its slipstream, such as Binge.
Record Club, which actually first opened its doors a couple of years ago, is continuing to grow and getting more attention as a potential Letterboxd for music enthusiasts. Music has been a major part of my life since my teens—across cassette tapes, CDs, MP3s, and streaming services—so I was keen to give it a try.
The snap verdict after a few days: It definitely gets a lot right, from the elegant look of the interface to the depth you can go down to when it comes to logging the details of your listening history. If you're looking for somewhere to share your music tastes and track your listening, it's worth a look.
Record Club is free to use with no ads, with a $4-per-month subscription available if you want to support the project. You get some extras such as better list management and the ability to choose preferred cover art in your library. You also get extensive personalized stats, and the ability to add personalized notes to releases.
When you first sign up for Record Club, you'll be given a quick tour of the interface and the features, and it's a well laid-out app on both web and mobile: There's lots of white space and text that's crisp and bold, and it's easy to read. Right away, you can see what's popular with other users, new releases coming up, and reviews that have been left on the platform.
Click through on any album, and you get a bunch of information about it. There's the track listing, plus details on how popular the album is on the Record Club network (its average rating and number of listeners, as well as reviews). You can see when it was originally released, and its total running time.
Everything is very clickable, searchable, and social. Follow a link to another user (you'll be given suggestions for who to follow as soon as you get started), and you get to see what they're currently listening to, their top five records, the reviews they've left on releases, and which upcoming releases they're planning to listen to.
Your own profile will be viewable by default: You can make changes to it by clicking on your avatar (top right), then choosing Settings > Profile. Switch to Preferences and you get to control some of the ways that Record Club works, and under Privacy you can choose what's shown on your profile, or make it private (so only people who know your username will be able to find you).
You can bring previous listening data over from Rate Your Music or Last.fm (which I'm a particular fan of), and imports from Spotify are apparently coming soon. Your Record Club data can be exported as and when needed, so you don't have to worry about your listening history staying trapped here forever.
The headers along the top of the web interface link to the key sections of the platform: Browse (for checking out new music), Community (for checking out what other listeners are up to), and Activity (for reviewing your own listening). It's really when you start to use the tracking and social features of Record Club that it really comes into its own though, and when the Letterboxd comparisons become most apt.
You can start sharing what you're listening to by heading to the Record Club home page and clicking Add release under Your rotation. Your rotation is where you log what you're currently listening to, and you can have five albums (or singles or EPs) here—if you try and add a sixth, something else will have to go.
Finding new music couldn't be any easier: Type the name of the artist and album, and it'll quickly pop up. You don't have to keep music titles in your rotation pile to tell Record Club that you've listened to them, as you can click the three dots on any cover art across the site and choose Listen to log it.
There's also Queue for music you want to listen to but haven't done yet, and the option to leave a rating or a review. I tend to leave ratings much more often than reviews, as they're easier: I'm usually agonizing over the right words to say in a review and the right tone to take, and then I end up not publishing it anyway.
It didn't take me long to log a lot of listening through the Record Club interface, as there were always related links and recommendations that took me where I wanted to go next. There's also the option to use tags if you want to be more specific in the way you're organizing your music collection.
If you head to your profile page by clicking your avatar (top right) and then Profile, you get your very own top five albums of all time to show off—just click Edit next to the relevant section to make your choices. One neat touch: If you click the three dots by the top five, you can save these picks as an image, so you can share them on other platforms as well.
I like making lists, too. Click Create up at the top of the web interface and you're able to choose List to start making some public or private lists. You can organize your favorite albums in a particular genre, a starter pack of songs for a specific band, the best double albums in history, or whatever it is you want to list.
We may earn a commission from links on this page.
As I've covered before, Tasker is the ultimate Android app for hacks, automations, and customizations. $3.99 from the Google Play Store, it gives you deep access into the settings and features of Android, so you can do more with Google's mobile operating system. It's actually not the easiest app to describe in just a few words—the best way to show off its capabilities is with examples of what it can do, like the ones I've collected here. They give you a small but carefully selected sample of ways to use Tasker, which you can use as inspiration for your own routines.
Admittedly, Tasker isn't the most straightforward app to get to grips with, but there is now an "easy mode" inside the app, as well as many preset tasks that you can load up without having to build anything yourself. There are also ideas available through the Tasker community online.
Sure, the default alarm sounds for Android are fine, but what if you could change the alarm sound based on what the weather is doing? You'd know from the moment you wake up what kind of day you were in for, in terms of the weather.
It's an example of the sort of tweaks Tasker can make that combine different features and data in useful ways. NotEnoughTech has a comprehensive tutorial you can work through in both video and written form to build this yourself, together with the necessary downloads.
Something else Tasker works well for upgrading basic Android features so you can get more out of them—like the rather rudimentary reminder system that you get through Gemini, Google Calendar, or Google Tasks.
Redditor Blitzdroids has built a Smart Reminders and Alarms project that you can download to do much, much more with reminders through Tasker. It's almost a mini-app in itself and it shows off what's possible.
The many customization features available for your reminders cover fonts, colors, sounds, and vibrations. You can control if and when reminders recur, whether or not they can override Do Not Disturb, and set bespoke snooze options too.
If there are wifi hotspots you use regularly, like at the local coffee shop, then you can use Tasker to have wifi automatically switch on when they're in range. It saves you doing it manually, and means you won't use any cellular data if you don't need to.
You can find instructions for this one over at MakeUseOf—it's pretty straightforward in Tasker terms. You need to set up a profile for each location, and then an associated task that automatically enables (and disables) wifi based on location.
Auto-rotate certainly has its uses—say, for watching videos—but there might be certain times when you don't want your apps automatically swinging around from portrait to landscape (or vice versa). Tasker is able to help here, too.
There's a preset project for this inside Tasker: Tap the three dots (top right), then TaskerNet, then find Auto-Rotation Apps via the search box at the top. Tap Import and then Import again to get the project on your phone.
As soon as the project is enabled, you'll be given a list of apps installed on your phone. The apps you select here will auto-rotate without prompting while the others won't, and you can go back into Tasker to edit the list at any time.
The best Tasker projects aren't always the most obvious ones: Take a look at ShakeWake, for example, which is designed to keep you alert when you're in danger of nodding off (during those long study or work sessions, maybe).
You get to set the necessary alarm interval for being pinged by your phone, and the pings can be accompanied by audible beeps or your phone's flashlight turning on. You can shake your phone to dismiss the alert, which is where the project name comes from.
You might remember all the way back to the days when Android used to have a ticker-style notification at the top, so the contents of your alerts would scroll along the top of the screen rather than popping up in a separate window. With Tasker, you can bring that functionality back.
This video shows how it works, and if you visit the Ticker project page on an Android device, you can load it on your phone. You can use it with some apps and not others, and it's able to display "now playing" information, too.
This is a Tasker project I was particularly happy to come across, because it affects me all the time: My phone often registers touchscreen inputs when it's in my pocket, I think because fingerprint or face detection input is somehow triggered.
That kind of problem should be solved by this Tasker idea, which uses the proximity sensor on your phone to detect that it's been pocketed, and switch to a custom lock screen. You can make certain exclusions too, to prevent accidental activation.
GPS access is useful for all kinds of apps—from checking the weather to tagging social media posts—but your phone's battery life can be significantly affected depending on whether GPS readings are being taken or not.
Tasker can help: This Reddit thread details the process of creating a simple Tasker project where opening a certain app turns GPS on, and closing it turns GPS off again. You could just have GPS enabled for Google Maps, for example.
As an added bonus, you know that multiple apps won't be accessing your location information—even if they have the necessary permissions inside Android, they won't be able to get at the data, as GPS won't be on.
Your mornings most likely follow the same sort of routine, except perhaps over the weekend. Tasker enables you to have the same app launch after the first unlock of the day, and even take actions on it (like pressing the play button).
From the Profiles screen, tap the three dots (top right), then TaskerNet. Search for "unlock" to find and import the project called Open App On First Unlock Of Day, and as soon as it's installed you can pick your app.
This is another project that shows Tasker can be capable in ways you might not have thought of, based on Android's out-of-the-box settings: You can have different apps open up based on which fingerprint is used to unlock the phone.
You could use this yourself—maybe thumb for the camera, and index finger for your messaging app. Or if there's someone else you trust enough to give them access to your phone, you could have a specific app open just for them.
It does take a little bit of effort to get this working, but it's not too taxing, and is worth the investment. There's an official video explaining how to do it from Tasker's creator, plus another tutorial on Reddit that outlines the steps.
An increasing number of powerful AI tools are now available to play around with, and they're becoming available to hackers and scammers as well as everyone else—which means a whole new set of attacks and dangers to watch out for. In the June Android Drop, Google's taking direct action against AI deepfake calls.
Here's how it works: A scammer sets up a call that appears to be from your mom. The caller ID is spoofed, and thanks to AI, it looks like your mom and it sounds like your mom. Before you can think of an excuse to not come home for Christmas, she's demanding $100 to cover an emergency with the plumbing.
With fake call detection, rolling out now in the Google Phone app on devices running Android 12 or later, that call should be flagged. You'll see a message on screen saying that your mom may not actually be your mom, and encouraging you to hang up. (Then you can call her from your end, to see if she really does need the money.)
Google says this detection works through the Rich Communication Services (RCS) messaging protocol, so both you and your trusted contacts will need to have RCS enabled in Google Messages. When someone calls you, a "silent confirmation signal" gets sent in the background from their phone to yours, verifying their phone and identifying them as one of your contacts.
If the call is from a deepfaked AI character, that signal will be missing, and the Phone app will tell you about it. The security feature is going to be on by default once the update rolls out, though you can turn it off through the settings screen in the Google Phone app (tap the menu button, top left, to find them).
It's a reminder of the highly realistic audio and video that AI is capable of producing now, which contributes to yearly financial losses from impersonation scams totaling $2.95 billion, according to figures quoted by Google. With this update applied, you should be a little bit safer.
There's more to talk about in the June Android Drop, as well, though fake call detection is the headline feature. As tends to be the case, some of these features have previously been announced for select handsets, and are now becoming more widely available.
That's the case with a new outfit search feature rolling out to Circle to Search: Previously available on Galaxy S26 and Pixel 10 devices, it's now coming to all phones running Android 14 or later that have Circle to Search installed. Highlight an outfit on screen, and you'll get search results for every part of that outfit together.
Then there's a wardrobe feature for Google Photos, announced last month and now heading to devices running Android 10 or later in the U.S., India, and Brazil. You get a new wardrobe section in the app, with items of clothing pulled from your photo library—you can browse these items by type, mix and match them together, and use AI to try on any combination of items.
More safety features are heading to the Personal Safety app for kids under 13, including medical information, emergency contacts, real-time location sharing, and car crash detection. Google has also confirmed that easier Android-to-iOS sharing, via AirDrop or QR code, is becoming more widely available.
There's an interesting update for Google Play Books: book insights. You get a "catch me up" option for summarizing the plot up to the current position (handy if you're returning to a book after an extended period), as well as the ability to ask questions about passages, themes, context, and characters.
Lastly, there are more combinations available in the Emoji Kitchen, Google's custom tool for combining emojis in a variety of fun ways—these emojis can be shared through the Gboard keyboard. With Android 17 now in beta testing as well, there should be more new features to look forward to in the coming months.
Claude is one of the AI platforms at the forefront of the generative AI revolution: It has become the go-to AI of choice for many coders, and is leading the push into agentic AI that can take actions for users, too. It's also really simple to use, and even opens up with a warm and familiar greeting ("back at it" when I last loaded it). However you use Claude, there are also numerous features and tools worth knowing about that can help you dive deeper into the AI's capabilities. Whether it's getting Claude to rummage through your emails, adjusting the style that Claude responds in, or putting Claude to work in your web browser, there is plenty to explore outside of the core prompt box.
Connectors enable Claude to attach itself to other apps, and there are a lot of them, including Spotify, Canva, Tripadvisor, and Uber. You can add a new connector to Claude by clicking on the + (plus) button in the lower left corner of the prompt box, then choosing Connectors > Add connector.
The Gmail connector is particularly useful. If you give Claude permission to access your inbox, the AI can carry out tasks like summarizing your daily messages, or identifying emails that need responding to. You could try "which email sender do I leave unread the most in my Gmail account?" for example, or "find emails in Gmail over the last 30 days that look as if they needed a response."
Claude can't generate AI images in the same way that ChatGPT and Gemini can, but it is able to produce diagrams, charts, and visualizations, and make them interactive. One of the best ways to make use of this is when you're learning about something: You could get Claude to make an interactive diagram of a volcano, for example, or a timeline of the music of the 1990s you can scroll through.
A prompt like "create an interactive visualization for me explaining how sound waves work" will show off what Claude can do. You'll get a simple animation in return showing how sound waves work, and sliders for adjusting the frequency and amplitude of the sound waves so you can see the differences it makes.
Claude usually does a good job of pulling up and synthesizing information from the web—you can make sure it searches online by clicking the + button in the prompt box, then "Web search"—but it doesn't always have a full idea about the sources that it's looking at.
A little prompt hacking can help here. If you're searching the web, specify that Claude should focus mainly on the most up-to-date information, from the most reputable website publishers (you can specify these, if you want). Adding a note to avoid rumor and speculation is also a good idea, and should mean Claude's responses are more reliable. Claude will embed web links in its answers for reference, so you can check if it did what it was told to.
Click the + button in the corner of the prompt box, then select Use style, and you'll see there are several different styles that you can get Claude to respond in: Learning, Concise, Explanatory, and Formal. There's also a Create & edit styles option for adding your own. You might want to create a style that gets Claude to reply with short, sharp, and to-the-point answers, or a style where Claude uses very simple language to explain complex topics. You could even set up a style where Claude talks like a pirate, if you want to.
Skills are another feature in Claude that can make a substantial difference to how much you get out of the AI. They're essentially sets of instructions that you can call on when needed—saving you from having to type out the instructions each time.
For example, you can use a skill to guide the tone and complexity of language to use when writing responses, or add instructions for how meeting notes should be reformatted and summarized—with specific headings or section lengths. A custom dictionary is another useful skill, where Claude can save definitions and bring them back again when needed.
Skills can be quite advanced and involve coding, but the easiest way to get started is to click the + button in the prompt box, then choose Skills > Add skill. You'll see there's actually a skill creator you can use, which lets you build skills through natural language prompts.
This one isn't about getting Claude to do something, but about stopping it, instead—specifically, blocking Claude from using your prompts to train its AI models. Click your user account name (lower left), then choose Settings > Privacy, and disable the feature labeled Help improve Claude. That way, you'll know your conversations are only used for your own purposes—not helping Anthropic improve its models.
Look up in the top right corner of any chat, and you'll a Share button. Click this, and you can create a public link for the chat that you can send to other people, or even post to the web. It's a read-only share though, so it's for disseminating information rather than collaborating on ideas. Anything you add to the conversation after you've shared it won't show up on the shared link, though it will if you unshare a chat and then share it again. As such, resharing links can make the experience a bit more collaborative: If others have input, you can ask Claude, allow it to respond, then reshare the chat link again. To manage the chats you've shared with the wider world, click your account name (lower left), then Settings > Privacy and Manage next to Shared chats.
Claude is able to create Word and Excel files from a single prompt, so you can get it to build a report on the state of the semiconductor industry or a chart showing off the achievements of your favorite sports team in just a few minutes. You can use this feature to create templates that you can fill out when needed. For example, ask Claude to "create an Excel file for family financial planning, with room to enter basic incomings and outgoings," and you get a fully formatted spreadsheet that you can then update.
Claude has a browser extension that you can install in Chrome or any Chromium-based browser (such as Microsoft Edge). You can then give it tasks for navigating websites, filling out forms, extracting data, and so on. The add-on will ask for approval for major actions, like making a purchase, and you can get it to ask for approval for every action by selecting Ask before acting in the prompt box that pops up when you launch the extension.
There's a lot the Claude extension can do. One helpful task the bot can carry out is looking at a list of hotels on a website and deciding which is the best fit for you, based on the criteria you give it (price, what the AI already knows about you, and any other information you want to add).
Sometimes you might not want your chats with Claude to be on the record, to remove any trace of them existing, and to stop Claude from referring back to them in the future. When you start a new chat, click the little ghost icon up in the top right corner to go incognito. To go back to normal, click the X in the top right corner. Your incognito chats are not used for training, but are retained by Anthropic for 30 days before being deleted.
It's difficult to get away from AI inside apps and on devices now, and if you're making a lot of use of these tools, you might be thinking about subscribing to a paid tier. While all the major AI chatbots offer free access, subscription plans offer more features and extended usage limits. Of course, there are a host of options to pick from—several big AI services, all with multiple plans at different price points. This guide should help you make some sense of the various offerings, and help you decide on the right one (or two) for you.
What's not included here is any assessment of which AI models are "better" than others. That can be difficult to measure, especially as models frequently change. Using the free tiers for these bots should give you a good idea of which AI models you like best, and which give the most relevant responses to your queries.
Google doesn't specify what the "standard limits" are for free Gemini usage, but it does say that AI Plus users get twice those limits, while AI Pro users get four times those limits. If you pay for AI Ultra, you get five times or 20 times the limits of AI Pro, depending on how much you're spending. The AI Pro and AI Ultra plans give you access to the latest AI models from Google.
Context windows go up with each tier, too, which is how much the AI can keep in its memory per conversation. The standard on free plans is 32K tokens (the blocks of info that AI bots work with), which add up to around 24,000 words. For AI Plus plans, that goes up to 128K tokens (about 96,000 words), and for all the other paid plans, you get one million tokens (about 750,000 words).
You can generate text, code, audio, images, and video with Gemini. It features a deep research mode, the ability to create custom AI bots with Gemini Gems, and scheduled actions. Most features are available on all the plans, but you'll find some exceptions. Video generation requires a paid plan, as does image editing. The new Gemini Spark AI agent is currently available only to AI Ultra subscribers.
Gemini makes sense if you already use a lot of Google apps (Gmail, Google Maps, Google Docs) and Android. After all, it's built right in. It's also worth noting that you get a host of extras included with these AI plans: more Google One storage at each tier, as well as YouTube Premium, Google Home Premium, and the new Google Health Premium (for AI Pro and AI Ultra plans).
Paid plans: AI Plus ($7.99 a month), AI Pro ($19.99 a month), AI Ultra (either $99.99 or $199.99 a month)
It's not particularly easy to figure out usage limits on the ChatGPT plans, but what we know for sure is that the Pro plans give you five times or 20 times the usage of Plus plans, for $100 or $200 a month, respectively. A Pro plan gives you unlimited messages with the latest models too, whereas on Go and Plus, you're limited to 160 every three hours. Also of note: The Go plan may display ads at times.
For context windows, again, OpenAI doesn't make it particularly clear. Free users get 16K tokens (roughly 12,000 words), which goes up to 32K tokens (24,000 words) for Plus users, but there's no mention of the Go plan. Pro users currently get 128K tokens (96,000 words) for GPT‑5.5 Instant, and 400K tokens (300,000 words) for GPT-5.5 Thinking—the best models ChatGPT has at the moment.
ChatGPT is able to generate text, code, and images for you. It has a deep research mode available to everyone, and support for custom AI bots (which it calls GPTs) on the Plus and Pro tiers. In general, paying more gets you more of everything: more messages, more uploads, more image generations, and access to the latest models first.
There's no wider ChatGPT suite of tools behind the AI chatbot, though it can connect up to a host of apps—including Photoshop, Figma, Spotify, Apple Music, Airtable, and plenty more. It also has its own experimental browser (ChatGPT Atlas), but it's very platform-agnostic when it comes to what it works best with.
Paid plans: Go ($8 a month), Plus ($20 a month), Pro ($100 or $200 a month)
You wouldn't really describe Claude's usage plans as straightforward, but obviously, the more money you pay a month, the more usage you get. Exactly how much usage that works out to "is affected by several factors, including the length and complexity of your conversations, the features you use, and which Claude model you're chatting with," according to the official documentation.
Context windows are the same on all three paid plan tiers: 200K tokens (about 150,000 words). What the context window is for free users isn't specified anywhere. Claude does let you pay as you go, to some extent, if you need extra AI usage on top of what you're already being charged for your subscription.
Coders frequently choose Claude for development, and, of course, it can also do general text output. It can't do images or video, though it can do basic visualizations (like charts and diagrams), and also offers the Canva-like Claude Design portal that will generate user interfaces, slideshows, and mock-ups for you. There's also a deep research tool here as well.
Like ChatGPT, and unlike Gemini, Claude isn't really tied to any particular ecosystem. However, it can connect to a multitude of other apps that you can then run through Claude, including Canva, Gmail, Slack, and Uber. There's also a Skills feature, where you can create customized instruction prompts for Claude to run again and again.
Paid plans: Pro ($20 a month), Max ($100 or $200 a month)
Copilot is unusual here, because (for individuals, rather than software developers) it's tied into Microsoft 365 (previously Microsoft Office) subscriptions. If you sign up for Word, Excel, and PowerPoint, then you get Copilot as well, at the two main pricing tiers.
While Microsoft isn't too precise about Copilot's usage limits, it does say that Personal subscribers get "higher than free" usage limits, and Premium subscribers get the "highest" limits. This applies to image generation too, while AI agents and audio creation are exclusive to the Premium tier. Microsoft doesn't say anything about context limits for individual users.
Copilot can create text and images, deploy a deep research mode, and connect to third-party services such as Gmail and Dropbox. There aren't quite as many of the AI bot features as you'll find elsewhere, like scheduled tasks or custom-made AIs for specific purposes, though there is a neat Study and Learn mode.
As with Gemini and Google, the primary reason to pick Copilot for your paid AI subscription is if you're already invested in Windows, the Edge browser, and Microsoft's office apps. Copilot integrates deeply into these pieces of software, and so is easier to call on if you need AI assistance.
Paid plans: Personal ($9.99 a month), Premium ($19.99 a month)
Perplexity models itself as more of an AI research tool and search engine, rather than an AI chatbot in the more general sense—like ChatGPT or Gemini. You can actually use Gemini, ChatGPT, and Claude models inside Perplexity, as well as Perplexity's own models (or use several in combination).
There's little official information from Perplexity about exact usage limits or context windows, though the more you pay, the more usage you'll get: The Max tier is described as offering the "highest usage and top performance", and gives you access to the "most advanced AI reasoning models".
Perplexity does offer tools like deep research and custom AI silos that it calls Spaces—complete with their own custom prompts and specific models. You can create images and videos with Perplexity, but it relies on third-party models for the task, and (anecdotally at least) usage limits seem to be quite tight.
The primary reason to sign up for a Perplexity subscription is if you find its focus on web search and information gathering useful. It also has its own browser, as well as some agentic tools that can carry out actions online for you. It's more focused than other AI chatbots, which may or may not be what you're looking for.
Paid plans: Pro ($20 a month), Max ($200 a year)
You generally open up Google Maps to get information rather than share it. Still, the wealth of data included in the app isn't as fixed as you might think: You can actually submit your own information and edits to Google Maps in numerous ways, from adding new places to correcting driving routes.
All of these submissions have to be reviewed by the Google team, so there's no guarantee that you'll see your changes actually appear on the map, but in theory, you'll be helping out the billions of other active Google Maps users. So, whether Google Maps is incorrectly advertising that your local bar does live music, or it's always sending people down the wrong road when they're trying to get to where you live, here's how to make changes directly from the app.
To add somewhere new in the Google Maps app, press and hold on the map at the right spot to drop a pin. If you swipe up on the pin's info card at the bottom of the screen, you can either Add a missing place or Add your business to Maps—if you choose the latter, Google Maps will ask you to fill out and verify a business profile, which you can then use to manage the place details in the future.
If you're adding a place you don't manage, you need to specify a name for it, and choose a category. Categories include Food & drink, Shopping, Services, Hotels and accommodation, and Outdoors & recreation, so you'll need to find the closest match.
You can submit a place with just a name, location, and category, but you have the option to add other details too, including a contact number, opening hours, and photos. Presumably, the more details you add, the better, though Google doesn't actually outline the approval process for getting new places added to the map.
Google does say that map data is constantly updated and reviewed, so it's probably automated and human processes working in tandem to determine whether or not your new place gets approved (for example, looking for other online evidence the place exists, or checking other user submissions).
You can also make changes to existing places on Google Maps, and indeed, that's actively Google actively encourages in its efforts to keep all of its map information up to date. If you notice that a park's opening times aren't actually the ones listed on Google Maps, you can submit the correct information.
Here's how to do that: With a location selected on the map, you can pull up the info card to see all of its details, and make changes. There are different screens where you can do this. On the Overview tab, for example, you can tap Suggest new hours under the opening hours, or Update location under the address if it's been positioned in the wrong place.
Swipe to the About tab, and you can see more specific details: For a restaurant, this might include whether or not it does delivery, or whether or not there's a parking lot. Tap Edit features and you can add your local knowledge in any of these categories.
Again, Google doesn't say how these edits are reviewed and approved, most likely in order to stop people from abusing the system. The app does indicate you may get an email about the status of one of your edits, and presumably, factors such as similar suggestions from other Google Maps users as well as your own history of contributions will be taken into account.
There are several other ways to get information updated on Google Maps. If you head to the Contribute tab, you'll find an Update road button: This lets you add missing roads, edit road names, indicate that a road is private or closed, or specify whether a road is one-way or two-way.
Giving feedback on directions is the only edit you have to make through Google Maps on the web rather than via one of the mobile apps. With directions on screen, click Details next to the directions on the left, then Send product feedback (bottom right). You'll be able to flag the steps that are wrong, and explain what's wrong with them.
A word of warning though: I've been trying to get Google Maps to give the right directions to my house for years (down the slightly longer, fully paved road rather than the slightly shorter, unpaved track). To date, the directions are still wrong, so you're not necessarily going to see your suggestions implemented.
There's also a general feedback form you can use for everything else to do with Google Maps. From inside the mobile app, tap your profile picture (top right), then choose Help and feedback > Send product feedback.
There are a flurry of AI-related announcements coming out of Google I/O 2026 today, but perhaps the most impressive is a new multimodal model called Gemini Omni. While it's launching as a video generator to begin with, it'll eventually be able to incorporate images and audio too, on both the input and output side.
The idea is you can remix different audio, images, and video into a completely new clip, via a custom prompt. Right now, you can only generate videos from text prompts and images within Gemini, so you're getting the added ability to combine audio clips and existing videos too when generating something new—multiple sources for input, and then an output that Google promises is better than ever in terms of realism and accuracy.
While image and audio generation is on the way, the ability to create videos is coming first, with a model called Gemini Omni Flash. The example Google gives is picking a few styles from images in your phone's gallery, and then applying them to an existing video: So if you wanted to, you could make a video of you in the real world look like a Pixar animation.
You can also edit your videos through "conversation," says Google. That conversation aspect will be familiar to anyone who already uses Gemini to make videos: You just explain what it is you want to see, and Omni takes care of it. You can use follow-up prompts to change something specific about the video, like an object or color, or to create your very own reshoots of the scene where the action changes.
You can also change the angle or the environment of a video—transporting yourself from a bedroom to a beach scene, perhaps. Google says you can take multiple turns to refine your videos, while still being able to get back to the original clip.
Google says Gemini Omni uses "an intuitive understanding of physics" together with "Gemini's knowledge of history, science, and cultural context" to make videos as realistic and as consistent as possible—though I'll have to try this out for myself to see if this all works as well as Google says it will.
Omni now comes with a better understanding of forces like gravity, kinetic energy, and fluid dynamics, so there should be less AI weirdness on show. As well as building scenes, Google says, Gemini Omni reasons about what should happen next.
AI videos can often collapse because they're trying to follow patterns from the vast number of videos in their training data, rather than follow the laws of physics. If a person disappears off-camera, they won't necessarily still be there when the camera pans back. Google claims Gemini Omni will show fewer issues like this.
To protect against deepfakes, Google is putting some limits on video creation. For now, you'll only be able to use your own voice and a digital avatar based on you to generate outputs. In addition, all videos will carry Google's invisible SynthID watermark that indicates the content is AI-generated.
Gemini Omni Flash is rolling out now in the Gemini app and Google Flow, for Google AI Plus, Pro, and Ultra subscribers. It's also going to be available for free in YouTube Shorts and the YouTube Create app later this week.
At the time of writing, there's no word on usage limits. At the moment, those on a Google AI Plus plan ($7.99 a month) can generate two videos a day using the Veo 3.1 Lite model. It remains to be seen how generous Google is with Gemini Omni generations—it looks like they take up a fair amount of AI processing power.
Lateo.net - Flux RSS en pagaille (pour en ajouter : @ moi) 