Feed aggregator

• Google’s Gemini AI assistant will start remembering your past chats
• You can opt out of the feature
• Gemini is also adding a new “Temporary Chat” mode for one-off conversations

Google Gemini will remember all those questions you ask on the app now, unless you ask it politely not to. The AI assistant will look at and reference your past chats as a way to personalize future ones, though only for Gemini 2.5 Pro for now.

With the memory function, Gemini could theoretically recall your favorite party themes, the YouTube channels you're into, or how you like to compose emails to your friends, without you having to remind it. Gemini's memory is not dissimilar to the memory feature offered by ChatGPT and other AI chatbots. And while ChatGPT can now connect to your Google account, Gemini has native access to Gmail, Calendar, and Google Docs.

The memory is on by default, so you'll have to make a slight effort if you'd rather Gemini not track your chats. You can turn the memory off or back on again in the settings menu under personal context, sliding the “Your past chats” option to off.

In addition, the “Gemini Apps Activity” section is now called “Keep Activity.” When that setting is active, some of your future uploaded files and photos may be used to train and improve Gemini and other Google services. You can opt out of that, too, if you wish.

Temp talks

(Image credit: Google)

As much as Google is eager to make sure Gemini can remember what you say to it, the company paired the rollout with its opposite, the new Temporary Chat. This incognito mode makes every conversation a one-off, and makes sure it isn't saved to your history, appears in the activity list, or sent to Google to train Gemini.

After a 72-hour hold for safety, they're they’re deleted completely, unless you submit explicit feedback, which will be processed and then discarded. Temporary Chat is designed for those moments when you want to ask something you’re not sure you’ll want remembered, whatever that might be.

The temporary chats might reassure people worried that Google just wants more of their data, but the company clearly hopes people find value in giving Gemini a memory. Google's goal is to encourage people to think of the AI as a long-term conversational partner, not just a tool you reintroduce yourself to every time you interact with it.

Regardless of any concerns, AI developers are very keen on memory, since, without it, AI assistants are just clever parrots. Technically, they still are just parroting data even with personalization, but they'll be less clever about it. Whether that’s exciting or unsettling depends on your appetite for intimacy with software.

You might also like

• I tried Google's new AI mode powered by Gemini, and it might be the end of Search as we know it
• Google previews AI Mode for search, taking on the likes of ChatGPT search and Perplexity
• Google Gemini's new model is the brainstorming AI partner you've been looking for

• Teenage Engineering has produced a new PC case
• The Computer-2 is a sequel to the Computer-1, but it's free
• It's made of plastic rather than aluminum, and unfortunately out of stock right now

Thinking of building a new computer? What if I told you that you could get the PC case for free - you probably wouldn't believe me, would you? Well, you can - with a couple of notable caveats, but this is still fundamentally a freebie.

The Verge reports that Teenage Engineering (more typically associated with nifty audio hardware) now has a sequel to its Computer-1 PC case. Can you guess what it's called? Yep: the Computer-2 case - in a stunning stroke of creativity.

The key difference is that, unlike the original case, which cost $ 149 in the US, the follow-up is free. Another notable change is that instead of aluminum, it's made of plastic, which is obviously a cost-saving measure.

In fact, it's a single sheet of semi-transparent plastic that folds into a PC chassis (using hinges and snap hooks). There are no screws needed, and the motherboard can simply be pushed and clicked into place, Teenage Engineering informs us.

It must be a (small) mini-ITX motherboard, mind, as this is a small form-factor (SFF) case for a compact PC build. It can fit an SFX power supply and a dual-slot graphics card (7 inches is the maximum length for the discrete GPU).

As well as the sheet of plastic that forms the chassis itself, you get the bits and pieces to go with it in terms of various power cables, fasteners, feet for the case to stand on, and so forth.

Analysis: What about those catches?

(Image credit: Teenage Engineering)

Okay, so coming back to the cost being a – literally – round figure of nothing - how on earth can this PC case be provided for free? Well, it can't, and Teenage Engineering is charging for the shipping to cover sending out the Computer-2 as you might guess (and likely a bit of that is to cover the cost of production). Of course, that's fair enough - it does mean you are getting the components themselves for free.

The other catch is that Computer-2 is already marked as sold out on the official website; you can only click to receive a notification when stock is back in. You can still examine the instructions, though, to get a better understanding of what construction involves, and exactly what you'll get.

By all accounts, the Computer-1 is a great case for a small PC build - and this new plastic take on the formula looks equally cool, especially given the price here (or lack of it). Anecdotally, I've seen a report of the shipping costing $16 (in the US) if you're wondering exactly what you might end up paying (when it's back in stock).

As for mulling over buying a cheap case in general, you can get some great products at the budget end of the market. Just be careful around airflow issues - especially if it's a gaming PC that you're building - and be sure to consult reviews to ensure you're not picking up a dud (and also our best PC cases buying guide will come in handy).

You might also like

• Built vs Bought: why prebuilt systems are always superior to custom gaming PCs
• Microsoft promises to crack one of the biggest problems with Windows 11: slow performance
• Microsoft has all but given up on Windows 11 SE – and it looks like the war against Chromebooks has been lost

OpenAI has started to roll out Gmail, Google Calendar, and Google Contacts integration into ChatGPT for Pro and Plus users today.

As a ChatGPT Plus subscriber, I was really looking forward to being able to connect my Gmail inbox to ChatGPT, as OpenAI CEO Sam Altman and the team demonstrated at the recent launch of ChatGPT-5.

Imagine my disappointment then, when, after going into ChatGPT’s Settings and connecting it all up to Google (a process which involves going through several permission screens), I found out that OpenAI is reserving the Gmail search in chat for ChatGPT Pro users.

Pro vs Plus

A ChatGPT Pro subscription costs $200 (£200 / AU$307 ) a month, compared to my $20 (£20 / AU$31) Plus subscription, just to use a feature that I get for free with my Google Gemini account!

Since Gemini and Gmail are both made by Google, I can understand that it must be a lot easier for Google to connect them, but I can’t quite understand the logic of putting the feature behind a $200 paywall for ChatGPT users.

I actually use Gemini to search my Gmail all the time - it’s much, much better than the standard search feature in Gmail because you can type in somewhat vague queries like “find me the email where I talk about trains and stamp collecting” and it does a great job of finding the exact email, even if you’ve no idea who it was from or when you sent it.

Having that feature in ChatGPT would have been great, but it seems it’s not going to happen for me anytime soon.

Deep Researching my Gmail

There is one caveat: On a ChatGPT Plus account, you can search Gmail through the Deep Research tool. Deep Research is OpenAI’s powerful research tool that’s meant to be used to research complex questions that require a full report to answer, not for simple things like searching your Gmail for a missing email.

For purely searching through your Gmail, I would say it works, but it is functionally useless: You click on Deep Research, then make sure that Gmail is turned on as one of the sources, then type in your query, and Deep Research ponderously goes off on its research process, which can take minutes.

It eventually comes back with good results, but nobody has got 5-10 minutes to wait for a simple Gmail search. I’d rather just go into Gmail and use the standard search bar and save myself minutes of time, or I’d use Gemini, which still seems like the best option for now.

You might also like

• OpenAI o3 checkmates Grok in a chess showdown, and it wasn't even close
• ChatGPT just gave me a hard truth that I probably needed to hear and I am shook
• ‘AI can teach you how to use itself, if you think to ask’ this Stanford professor’s ChatGPT prompt has changed the way I use AI for good - here’s how to try it

The whole idea of an AI phone is a rather new one, but I’d argue that Google has been both the driver and frontrunner of phones that put artificial intelligence at their center.

This started with the Google Pixel 6 and its Tensor chip, which was built around delivering strong neural network machine-learning performance rather than the best CPU and graphics processing speeds. Then, with the Pixel 8, we saw Google really embrace AI features, especially those with generative capabilities like the Magic Editor.

While some might argue that the Samsung Galaxy S24 was the first ‘AI phone’, I’d say the Pixel 8 was the one that got the ball rolling, especially as Samsung used Google tech to underpin some Galaxy AI features.

This was all built upon with the Google Pixel 9 and its stablemates, which all felt like they put AI at the heart of the Pixel experience, rather than just offer tools on top of a standard smartphone user interface.

But with the Samsung Galaxy S25 and Galaxy Z Fold 7 landing with improved AI features, and Apple Intelligence starting to get up to speed and offer a solid suite of features on compatible iPhones, as well as some synchronicity with Macs, Google’s AI phone crown could be up for grabs.

So I feel that for Google to keep ahead and continue to blaze a trail for practical, and hopefully safe, consumer use, it'll need to bring more to the table with the Pixel 10.

Accelerated AI

(Image credit: Future / Google)

But what would that look like? Well, that's a good question.

In some ways, Google already offers up a good suite of AI-powered tools, from smart call handling and document summaries to real-time translation and photo editing tools that can rework entire shots. And a lot of these are easy to access and use, something which helped make the Google Pixel 9 Pro our phone of 2024.

However, there’s room for more. Ideally, I’d like to see some of these smart AI tools work across the whole of the Pixel Launcher experience.

For example, I’d like to use Gemini to summarize what I’ve got on my Pixel phone, say, telling me apps that I’ve not used for ages or games that I’ve not played and may have forgotten about.

And I’d be very keen to have the summarization tools built out to provide me with snippets of information in third-party apps, say to quickly summarize a chapter of an ebook so I can easily pick up where I’ve left off without reading back over stuff for a refresher.

More than anything else, I’d love to have audio readouts of all kinds of text in a voice as close to natural language speaking as possible. I listen to a lot of podcasts, especially when I'm commuting or cooking, but this does mean I can fall behind on a backlog of reading.

To tackle this, I’d love to be able to ask a system-level AI to read aloud a magazine feature either from a digital source or scanned in via a phone’s camera. I’ve stumbled across some AI tools that can sort of do this, but none deliver what I’m looking for and in a seamless way that AI luminaries tend to tout.

This isn’t a pipedream, as Google already has NotebookLM with its Audio Overviews that can make a podcast out of various inputs. So if this tech could be baked into the next-generation Pixel Launcher or the rumored Pixel 10 phones, it would be a boon for me and likely other people looking for aural satisfaction.

Ultimately, while I appreciate the creative opportunities of some AI tools in terms of generating images or deeply manipulating photos, I’d like AI in general to be more about making my daily life easier and helping me better digest the almost overwhelming amount of content there is to consume in the virtual and real world.

I have faith that Google is one of the companies best placed to do this, even with my nervousness about how much the search giant’s algorithms already influence the dissemination of information.

But I do think it needs to use each new generation of Pixel phone to keep pushing the envelope, while the likes of Apple and Samsung appeal to the everyday phone users, not just to keep the AI phone crown, but also to push AI innovation forward in a way that’s genuinely useful and productive for people.

You might also like

• The Google Pixel 10 could have the iPhone 17 beat even before Apple’s phone arrives – here’s how
• Will the Google Pixel 9's price drop when the Pixel 10 launches?
• 5 ways the Google Pixel 10 Pro can get ahead of the iPhone 17 Pro and Galaxy S26 Ultra

Reuters found that Meta's AI guidelines and standards are troublesome in multiple ways.

With a wild card invitation, the tennis legend is making history in her return to the U.S. Open later this month as the oldest singles player to take the court in more than 40 years.

(Image credit: Scott Taetsch)

Here are hints and the answers for the NYT Connections: Sports Edition puzzle for Aug. 15, No. 326

Here are hints and answers for the NYT Strands puzzle for Aug. 15, No. 530.