NVIDIA starts offering a 12GB version of the 5070 for laptops

NVIDIA is releasing a new variant of its 5070 GPU for laptops. Nestled in a blog post about the latest version of its Game Ready Drivers, the company notes its partners will soon start selling 5070 laptops with 12GB of VRAM, alongside the 8GB model that NVIDIA has offered since the launch of the 50-series. 

"Demand for GeForce RTX remains strong, and memory supply is contrastrained. In order to maximize memory availability, we are releasing the GeForce RTX 5070 Laptop GPU 12GB configuration with 24Gb G7 memory. This gives our partners access to an additional pool of memory to complement the 16Gb G7 supply that currently ships with most GeForce GPUs," NVIDIA said. 

The first 12GB 5070-equipped laptops are slated to start shipping sometime in June, with manufacturers like ASUS, Lenovo and MSI likely to offer the video card as an option in some of their models. NVIDIA has yet to confirm pricing, but outlets like NotebookCheck are reporting that 12GB 5070 laptops could cost as much as their 5070 Ti counterparts. Right now, a 5070 Ti-equipped PC like the Acer Predator Helios Neo 16S AI can set you back as much $2,650, depending on the amount of RAM on offer. New 12GB 5070 laptops likely won't cost as much, given manufacturers will probably configure them with less RAM.

NVIDIA has yet to share the full spec list for the 12GB 5070, but as the company notes, it’s using 24Gb G7 memory, instead of 16GB G7 memory, for the new model. The two memory types are made using different manufacturing processes. The former uses 3GB memory modules, while the latter uses 2GB ones. Either way the company is tapping into a different supply of memory that, in recent months, Samsung and Micron have managed to produce more consistently at scale. That said, unless NVIDIA has redesigned the 5070 to equip it with a wider 192-bit bus interface, which seems unlikely in this case, the new model won't be able to access that additional memory as fast as the 5070 Ti and other models above it in NVIDIA's stack. For most games that shouldn't matter too much, but it does mean the new model isn't quite the upgrade it seems if you just look at the amount of raw VRAM.             

 

This article originally appeared on Engadget at https://www.engadget.com/computing/laptops/nvidia-starts-offering-a-12gb-version-of-the-5070-for-laptops-180057515.html?src=rss

What you need to know as Elon Musk’s lawsuit against Sam Altman begins

In a few short days, jury selection will begin in the long-awaited Musk v. Altman case. At the end of that process, an Oakland federal court will task nine regular people with deciding if OpenAI defrauded Elon Musk when it announced, and recently completed, its reorganization to become a more traditional for-profit business. More than just being the venue where two billionaires will air their grievances against one another in public, the trial has the potential to reshape the AI industry.

Musk first sued OpenAI in 2024, but the seed of the dispute was planted when Sam Altman emailed the billionaire on the evening of May 25, 2015. “Been thinking a lot about whether it’s possible to stop humanity from developing AI. I think the answer is most definitely not,” Altman wrote at the time. “If it’s going to happen anyway, it seems like it would be good for someone other than Google to do it first. Any thoughts on whether it would be good for [Y Combinator] to start a Manhattan Project for AI?”

“Probably worth a conversation,” Musk responded a couple of hours later. That same year, OpenAI announced itself to the world, with Altman and Musk as co-chairs of the new joint venture. “OpenAI is a nonprofit artificial intelligence research company. Our goal is to advance digital intelligence in the way that is mostly likely to benefit humanity as a whole, unconstrained by a need to generate financial return. Since our research is free from financial obligations, we can better focus on a positive human impact.”

If we’re to believe OpenAI’s telling of the events that followed, by 2017, almost everyone at the company, including Musk, agreed that a for-profit entity “had to be part of the next phase for OpenAI,” due to the enormous amount of investment needed to pursue its original mission. At some point before Musk left OpenAI’s board of directors in February 2018, OpenAI claims he demanded full control of the company, with the intent to eventually merge it with Tesla.

Following Musk’s departure, OpenAI created its for-profit arm in 2019, which at the time was organized under a “capped-profit” structure designed to limit investor returns to 100x, with any excess windfalls flowing to the company’s nonprofit. The idea being that if OpenAI achieved artificial general intelligence, its nonprofit would be the greatest beneficiary. However, after the success of ChatGPT in 2022, that structure became problematic for OpenAI as the company sought to raise ever more capital, and as part of its $6.6 billion funding round in October 2024, it reportedly agreed to a less-than-two-year deadline to free its for-profit from control of the nonprofit.

“At the heart of this trial is that OpenAI began as a non-profit organization, and then decided that it needed to be a for-profit organization in order to raise the enormous sums of money it needed to develop the technology it wanted to create,” explains Professor Michael Dorff, executive director of the Lowell Milken Institute for Business Law and Policy at UCLA. “That is a very troublesome transition under the law.”

Earlier this year, following protracted negotiations with Microsoft (the for-profit’s largest investor) and the state attorneys general of California and Delaware, OpenAI announced the successful reorganization of its corporate structure. As things stand, the for-profit is now a public benefit corporation, making it more appealing to investors looking for an uncomplicated return structure. Meanwhile, the nonprofit — now known as the OpenAI Foundation — holds equity in the for-profit arm, a stake valued at $130 billion at the time the agreement was announced. 

At the end of last year, Musk filed an injunction to prevent the reorganization from going through but failed. As an early donor to OpenAI, Musk will not see a single cent of money come his way when the company holds an initial public offering, on account of the fact donations are made with no expectation of any return. Musk has therefore argued OpenAI’s founding group, including CEO Sam Altman and President Greg Brockman, defrauded him as a donor.

Determining the exact amount Musk contributed to OpenAI was an early question during pre-trial discovery. You see, Musk has greatly exaggerated his monetary contributions. As recently as March 2023, the billionaire regularly claimed he had donated about $100 million to OpenAI. He later cut that estimate by half, telling CNBC in May 2023: “I’m not sure the exact number but it’s some number on the order of $50 million.” In recent court filings, that number was again revisited to $38 million, and it’s the number that currently stands.

In his original complaint, Musk’s legal team tried to “throw the kitchen sink” at OpenAI, says Professor Dorff. In subsequent filings, Musk’s lawyers narrowed down their client’s desired set of outcomes to a handful of remedies. Should the jury rule in his favor, Musk has requested the court force Altman and Brockman to step down, and for OpenAI to restructure as “a bona fide public charity that operates as the nonprofit it was intended to be, consistent with its founding charter and mission.” He's also made the highly unusual request that any monetary damages which would be awarded to him in the verdict be redirected to OpenAI's own nonprofit arm.

According to Professor Dorff, it’s highly unlikely Musk will be able to undo OpenAI’s reorganization. For one, District Judge Yvonne Gonzalez Rogers has already signaled her reluctance to do just that — and it’s her, not the jury, who will get to decide if that’s an appropriate remedy. Effectively, Musk is asking the judge to “unscramble the eggs” of a complicated corporate restructuring.

“There was a moment where that might have been possible, when the attorneys general of Delaware and California intervened and came to the current compromise,” explains Dorff. “Whether you agree or disagree with what the AGs decided to do, I think it's unlikely the court will feel it's appropriate to undo that compromise because of all the high government officials involved who, in theory, had all of the right incentives.” When Musk filed his request for a preliminary injunction to stop OpenAI’s conversion to a for-profit company, the judge said the request was “extraordinary and rarely granted.” The fact Musk is deeply involved with OpenAI's competitor xAI “may also weigh heavily on the judge's mind,” Droff adds.

Far more uncertain is how Musk’s other demands could play out, since the jury will decide if OpenAI is guilty of defrauding him. According to Dorff, most high-stakes business cases end with the two sides settling because of the risk of involving a jury in the outcome. “I just don’t see that happening here given the tenor of the dispute,” he says. “It seems unlikely either side will settle.”

If the case does end in a jury decision, it will then be up to those nine people, with guidance from the judge, to decide on monetary damages. “That will be very difficult to figure out because there is a maximalist version of this, and a minimalist version of this. They’re very different numbers and the result could be anywhere in between the two,” says Dorff. Musk’s legal team is seeking a disgorgement of between $65.5 billion and $109.43 billion from OpenAI (and between $13.3 billion and $25.06 billion from Microsoft, which is a co-defendant in the case). In a worse case scenario, Professor Dorff suggests Altman might lose the confidence of OpenAI’s board, costing him his position as CEO. He might even be forced to write some checks to settle the disgorgements.

Dorff suspects OpenAI “would love” the minimalist version where Musk is rewarded his $38 million donation back. Should some other disgruntled donors emerge to sue OpenAI for fraud, the Musk v. Altman case would make it easier to litigate those cases, given “the map has been drawn as to which legal claims are likely to succeed,” says Dorff. However, those would amount to “traffic tickets” for OpenAI.

Whatever happens next, it should be an eventful trial. With public testimonies from Microsoft CEO Satya Nadella, former OpenAI board member and Musk confidant Shivon Zilis and even Altman himself a likelihood, we'll at the very least be treated to a wealth of formerly private communications — and some new piece of vocabulary — between some of the richest people in the tech space. 

This article originally appeared on Engadget at https://www.engadget.com/ai/what-you-need-to-know-as-elon-musks-lawsuit-against-sam-altman-begins-191500726.html?src=rss

What you need to know as Elon Musk’s lawsuit against Sam Altman begins

In a few short days, jury selection will begin in the long-awaited Musk v. Altman case. At the end of that process, an Oakland federal court will task nine regular people with deciding if OpenAI defrauded Elon Musk when it announced, and recently completed, its reorganization to become a more traditional for-profit business. More than just being the venue where two billionaires will air their grievances against one another in public, the trial has the potential to reshape the AI industry.

Musk first sued OpenAI in 2024, but the seed of the dispute was planted when Sam Altman emailed the billionaire on the evening of May 25, 2015. “Been thinking a lot about whether it’s possible to stop humanity from developing AI. I think the answer is most definitely not,” Altman wrote at the time. “If it’s going to happen anyway, it seems like it would be good for someone other than Google to do it first. Any thoughts on whether it would be good for [Y Combinator] to start a Manhattan Project for AI?”

“Probably worth a conversation,” Musk responded a couple of hours later. That same year, OpenAI announced itself to the world, with Altman and Musk as co-chairs of the new joint venture. “OpenAI is a nonprofit artificial intelligence research company. Our goal is to advance digital intelligence in the way that is mostly likely to benefit humanity as a whole, unconstrained by a need to generate financial return. Since our research is free from financial obligations, we can better focus on a positive human impact.”

If we’re to believe OpenAI’s telling of the events that followed, by 2017, almost everyone at the company, including Musk, agreed that a for-profit entity “had to be part of the next phase for OpenAI,” due to the enormous amount of investment needed to pursue its original mission. At some point before Musk left OpenAI’s board of directors in February 2018, OpenAI claims he demanded full control of the company, with the intent to eventually merge it with Tesla.

Following Musk’s departure, OpenAI created its for-profit arm in 2019, which at the time was organized under a “capped-profit” structure designed to limit investor returns to 100x, with any excess windfalls flowing to the company’s nonprofit. The idea being that if OpenAI achieved artificial general intelligence, its nonprofit would be the greatest beneficiary. However, after the success of ChatGPT in 2022, that structure became problematic for OpenAI as the company sought to raise ever more capital, and as part of its $6.6 billion funding round in October 2024, it reportedly agreed to a less-than-two-year deadline to free its for-profit from control of the nonprofit.

“At the heart of this trial is that OpenAI began as a non-profit organization, and then decided that it needed to be a for-profit organization in order to raise the enormous sums of money it needed to develop the technology it wanted to create,” explains Professor Michael Dorff, executive director of the Lowell Milken Institute for Business Law and Policy at UCLA. “That is a very troublesome transition under the law.”

Earlier this year, following protracted negotiations with Microsoft (the for-profit’s largest investor) and the state attorneys general of California and Delaware, OpenAI announced the successful reorganization of its corporate structure. As things stand, the for-profit is now a public benefit corporation, making it more appealing to investors looking for an uncomplicated return structure. Meanwhile, the nonprofit — now known as the OpenAI Foundation — holds equity in the for-profit arm, a stake valued at $130 billion at the time the agreement was announced. 

At the end of last year, Musk filed an injunction to prevent the reorganization from going through but failed. As an early donor to OpenAI, Musk will not see a single cent of money come his way when the company holds an initial public offering, on account of the fact donations are made with no expectation of any return. Musk has therefore argued OpenAI’s founding group, including CEO Sam Altman and President Greg Brockman, defrauded him as a donor.

Determining the exact amount Musk contributed to OpenAI was an early question during pre-trial discovery. You see, Musk has greatly exaggerated his monetary contributions. As recently as March 2023, the billionaire regularly claimed he had donated about $100 million to OpenAI. He later cut that estimate by half, telling CNBC in May 2023: “I’m not sure the exact number but it’s some number on the order of $50 million.” In recent court filings, that number was again revisited to $38 million, and it’s the number that currently stands.

In his original complaint, Musk’s legal team tried to “throw the kitchen sink” at OpenAI, says Professor Dorff. In subsequent filings, Musk’s lawyers narrowed down their client’s desired set of outcomes to a handful of remedies. Should the jury rule in his favor, Musk has requested the court force Altman and Brockman to step down, and for OpenAI to restructure as “a bona fide public charity that operates as the nonprofit it was intended to be, consistent with its founding charter and mission.” He's also made the highly unusual request that any monetary damages which would be awarded to him in the verdict be redirected to OpenAI's own nonprofit arm.

According to Professor Dorff, it’s highly unlikely Musk will be able to undo OpenAI’s reorganization. For one, District Judge Yvonne Gonzalez Rogers has already signaled her reluctance to do just that — and it’s her, not the jury, who will get to decide if that’s an appropriate remedy. Effectively, Musk is asking the judge to “unscramble the eggs” of a complicated corporate restructuring.

“There was a moment where that might have been possible, when the attorneys general of Delaware and California intervened and came to the current compromise,” explains Dorff. “Whether you agree or disagree with what the AGs decided to do, I think it's unlikely the court will feel it's appropriate to undo that compromise because of all the high government officials involved who, in theory, had all of the right incentives.” When Musk filed his request for a preliminary injunction to stop OpenAI’s conversion to a for-profit company, the judge said the request was “extraordinary and rarely granted.” The fact Musk is deeply involved with OpenAI's competitor xAI “may also weigh heavily on the judge's mind,” Droff adds.

Far more uncertain is how Musk’s other demands could play out, since the jury will decide if OpenAI is guilty of defrauding him. According to Dorff, most high-stakes business cases end with the two sides settling because of the risk of involving a jury in the outcome. “I just don’t see that happening here given the tenor of the dispute,” he says. “It seems unlikely either side will settle.”

If the case does end in a jury decision, it will then be up to those nine people, with guidance from the judge, to decide on monetary damages. “That will be very difficult to figure out because there is a maximalist version of this, and a minimalist version of this. They’re very different numbers and the result could be anywhere in between two,” says Dorff. Musk’s legal team is seeking a disgorgement of between $65.5 billion and $109.43 billion from OpenAI (and between $13.3 billion and $25.06 billion from Microsoft, which is a co-defendant in the case). In a worse case scenario, Professor Dorff suggests Altman might lose the confidence of OpenAI’s board, costing him his position as CEO. He might even be forced to write some checks to settle the disgorgements.

Dorff suspects OpenAI “would love” the minimalist version where Musk is rewarded his $38 million donation back (and it ends up with the company’s non-profit). Should some other disgruntled donors emerge to sue OpenAI for fraud, the Musk v. Altman case would make it easier to litigate those cases, given “the map has been drawn as to which legal claims are likely to succeed,” says Dorff. However, those would amount to “traffic tickets” for OpenAI.

Whatever happens next, it should be an eventful trial. With public testimonies from Microsoft CEO Satya Nadella, former OpenAI board member and Musk confidant Shivon Zilis and even Altman himself a likelihood, we'll at the very least be treated to a wealth of formerly private communications — and some new piece of vocabulary — between some of the richest people in the tech space. 

This article originally appeared on Engadget at https://www.engadget.com/ai/what-you-need-to-know-as-elon-musks-lawsuit-against-sam-altman-begins-191500726.html?src=rss

ChatGPT Images 2.0 is better at rendering non-Latin text

A little more than a year after OpenAI gave ChatGPT users the option to create images and designs directly from its chatbot, it's now releasing ChatGPT Images 2.0. OpenAI describes the new system as a “step change” for image generation models, particularly when it comes to the tool’s ability to follow instructions in detail, render dense text and place and relate objects in a scene. For the first time, OpenAI has also built an image model with reasoning capabilities, giving the system the ability to do things like search the web and verify its outputs. According to the company, those capabilities should translate to a tool that's more reliable when accuracy, consistency and visual cohesion are essential. 

An example of ChatGPT's new non-Latin rendering abilities.
An example of ChatGPT's new non-Latin rendering abilities.
OpenAI

OpenAI says it has also put in a lot of work to make Images 2.0 better at understanding and rendering non-Latin text, with "significant gains" when it comes to the model's ability to handle Japanese, Korean, Chinese, Hindi and Bengali. At the same time, the company claims the new model is better at faithfully recreating the specific characteristics of different visual languages. On this point, OpenAI says that makes Images 2.0 more useful for tasks like game prototyping and storyboarding. Outside of those features, the new model is more flexible when it comes to aspect ratios, allowing it to generate images that are as wide as 3:1 and as tall as 1:3. It can also produce designs at resolutions of up to 2K, and even generate up to eight outputs in one go. 

A tortoiseshell cat in the style of Pokemon's third generation of games.
A tortoiseshell cat in the style of Pokemon's third generation of games.
ChatGPT

I got a chance to preview Images 2.0 ahead of its public release. For my first prompt, I asked ChatGPT to generate an image of a tortoiseshell cat in the pixel art style of Pokémon's third generation. I thought this would be a good test because AI models typically struggle with pixel art, and the Game Boy Advance Pokémon games are iconic for their art style, so much so that if ChatGPT merely approximated that style, it wouldn't do. The result is the image you see above, and I think ChatGPT did a commendable job there. I then tasked the new model with converting that image into a transparent PNG. For one last test, I asked ChatGPT to create a four-page manga about my cat enjoying a sunny day by an idyllic city stream. 

Notice how the cat isn't render exactly like the one above it.
Notice how the cat isn't render exactly like the one above it.
ChatGPT

Of those three tests, ChatGPT spent the most time on the second one and the output there was slightly different from the first image it generated, which I felt deviated from my prompt. Still, it managed to generate a proper transparent image, which is something other image models can struggle to do properly. Once more people have a chance to put the model through its paces, we’ll have a better idea of how it compares to Google’s Nano Banana 2, and where OpenAI can make additional improvements.

A manga generated by ChatGPT about a cat enjoying a sunny day.
A manga generated by ChatGPT about a cat enjoying a sunny day.
ChatGPT

Images 2.0 is available starting today for all ChatGPT users, including those on the company's Free and Go tiers. Plus and Pro subscribers get access to more advanced outputs. OpenAI is also making the model available through its API service and Codex coding app, which just last week it updated to offer built-in image generation. Notably, Images 2.0 arrives just days after Anthropic waded into the visual design market with its own design assistant.  

This article originally appeared on Engadget at https://www.engadget.com/ai/chatgpt-images-20-is-better-at-rendering-non-latin-text-190000153.html?src=rss

Google brings Gemini in Chrome to users in Asia and the Pacific

After debuting in the US, Gemini in Chrome is making its way to more markets. Starting today, Google is rolling out Chrome's built-in chatbot to users in Asia and the Pacific, including Australia, Indonesia, Japan, the Philippines, Singapore, South Korea and Vietnam. The expansion comes after Google earlier this year made Gemini in Chrome available to people in Canada, India and New Zealand

With the exception of Japan, where Google isn't making the new suite available on iOS just yet, everyone else in the countries mentioned above can access Gemini in Chrome through Chrome's desktop browser, and the app on their iPhone or iPad. To get started, just tap the "Ask Gemini" icon at the top right of the screen. It will open a new sidebar Google introduced at the start of the year where you can chat with Gemini across every open tab. From there, you can also access Google's in-house image generator, Nano Banana 2. As you would expect, the suite offers integrations with Google's other apps, allowing you, for instance, to add events to Calendar without leaving the interface. 

If you don't want to use Gemini, you can right click on the shortcut to unpin it from the top of the interface.    

Update 7:43PM ET: This article has been updated to reflect the expansion includes the entire Asia-Pacific region.

This article originally appeared on Engadget at https://www.engadget.com/ai/google-brings-gemini-in-chrome-to-users-in-asia-and-the-pacific-220000698.html?src=rss

Anthropic now has a design assistant too

In hindsight, I suppose it was only a matter of time after Anthropic made Claude capable of generating charts and diagrams that the company would then begin offering a more robust image editor. Now, a little more than a month after that release, Anthropic has announced Claude Design, a new research preview that allows subscribers to use Claude to generate designs, prototypes, slides and more.  

"Claude Design gives designers room to explore widely and everyone else a way to produce visual work," Anthropic says of its newest product. As with its previous forays into image generation, the company isn't calling this, well, an image generator. Instead, Anthropic describes Opus 4.7, the system powering the app, as its most capable vision model to date. In other words, you won't be using Claude Design to whip up a picture of a cat in space eating a lasagna.  

As you might expect, every project in Claude Design starts with a prompt. From there, Anthropic notes users can refine Claude's outputs through conversation, inline comments and direct edits. Like Adobe's recently announced AI assistant, Claude will also generate custom sliders that correspond to specific elements in a design, which the user can push and pull to modify those elements. For instance, in the screenshot below, you can see how Claude has tweaked the interface to allow the user to adjust the glow and density of arcs it used to illustrate a connected network.

Claude Design will generate custom sliders you can use to adjust specific visual elements.
Anthropic

Anthropic has also built an onboarding process that allows Claude to build an internal visual language after reading your organization's codebase and existing design documents. "Every project after that uses your colors, typography, and comments automatically," according to the company. Outside of text prompts, there's also support for image and document uploads, and Anthropic has even included a web capture tool so enterprise customers can snapshot elements from their company's website. There's also built-in sharing, and you can export a design directly to Claude Code. In the coming weeks, Anthropic has promised to make it easier to build integrations with its new app. 

Claude Design arrives in the same week that both Adobe and Canva released their own visual AI assistants. If Anthropic is preparing to eat Canva’s lunch, it's doing so in a strange way given that you can export your Claude Design projects to Canva. If you want to try the new app for yourself, it's available as part of Anthropic's Pro, Max, Team and Enterprise subscriptions, with usage running up against your usage limits.

This article originally appeared on Engadget at https://www.engadget.com/ai/anthropic-now-has-a-design-assistant-too-150000903.html?src=rss

Perplexity brings its Personal Computer AI assistant to Mac

Perplexity has just released Personal Computer. The software, which is available starting today for Mac, builds on the multi-model orchestration capabilities the company debuted with Perplexity Computer at the end of February. Like Claude Cowork (and, as of today, OpenAI Codex too), it's a suite of computer use agents that can work with your files, apps, connectors and the web to complete complex and "even continuous workflows." 

Perplexity suggests a few different use cases for Personal Computer, starting with the obvious. “You can ask Personal Computer to read your to-do list,” the company states. “In fact, you can ask it to DO your to-do list." It explains you can open the Notes app on your Mac, ask Personal Computer for help and the system will reason how to best assist you. In the process of tackling that task, it can work across all your files, as well as apps like Apple Messages. When needed, it will also employ multiple agents to complete a request. Like Anthropic did with Claude Cowork, Perplexity says you can also use its software to organize messy folders so files feature sensible names and there's an easy-to-understand structure to everything.

You can prompt Personal Computer with your voice, and you can even initiate and manage tasks from your phone. Perplexity says the app creates files in a secure sandbox, and any actions it takes are auditable and reversible. "A system that acts on your behalf needs to be useful and legible. It should feel like a team you manage, not a rogue employee with keys to your most important data," the company said.   

Personal Computer for Mac is available starting today, beginning with Max subscribers. Perplexity said it would bring the app to its other users soon, prioritizing those who joined the waitlist for the experience. 

This article originally appeared on Engadget at https://www.engadget.com/ai/perplexity-brings-its-personal-computer-ai-assistant-to-mac-202045969.html?src=rss

OpenAI’s latest Codex update builds the groundwork for its upcoming super app

Last month, following reporting from The Wall Street Journal, OpenAI confirmed it was working on a desktop super app that would combine ChatGPT, its Codex coding agent and Atlas web browser into one cohesive experience. OpenAI is not releasing that application today. Instead, it's pushing out a major update to Codex that significantly expands what that software can do. However, the new release offers a glimpse of what OpenAI hopes to build with its latest effort.  

"We're building the super app out in the open," said Thibault Sottiaux, the head of Codex, during a press briefing held by OpenAI. "This release is about developers. In the future, we will broaden it up to a wider audience." Until then, the latest version of Codex offers developers multi-purpose AI agents that can work across a "larger surface area," while being more proactive. In practice, that translates to a host of new capabilities, starting with computer use. 

The agents inside of Codex can interact with other apps on your PC. When prompting one of OpenAI's models, you can name a specific program or let it determine the best application for the job. Computer use is available in competing apps like Claude Cowork, but where OpenAI believes Codex offers an edge in that department is in the "secret sauce" it built to allow an agent to run an app without bogging down your entire system, so the two of you can work in tandem. At the same time, OpenAI is releasing 111 new plugins for Codex that combine skills, app integrations and model context protocol server connections to give Codex more ways to gather context and use the tools developers depend on for their work.

The company has also added a built-in browser, with a commenting system that allows you to prompt Codex to make tweaks to specific parts of a webpage or web app you're building. In the demo OpenAI showed, one member of the Codex team used this tool to instruct Codex to change the margins on a graph so that the y axis wasn't cut off. Complementing this is built-in image generation. Codex can use gpt-image-1.5 to create product concepts, mockups, frontend designs and even assets for simple games. It also allows Codex to use screenshots to verify it's on the right track with a user request.   

With today's update, OpenAI is also previewing a pair of memory features. The first allows Codex to recall context from previous tasks to inform how it goes about future prompts. According to OpenAI, with time, this will allow Codex to complete requests faster and to a higher standard. The app will also use the context it's gathered to suggest proactive actions. For example, at the start of your day, it might suggest you respond to a comment a coworker left on a Google Doc draft you wrote. 

If you want to try the updated Codex for yourself, OpenAI is starting to roll out the new version to desktop app users who are logged in with their ChatGPT account. Computer use is available to macOS users first, with availability for people in the EU and UK to follow soon. Similarly, Brits and Europeans will need to wait to try the memory features OpenAI has built into Codex.  

This article originally appeared on Engadget at https://www.engadget.com/ai/openais-latest-codex-update-builds-the-groundwork-for-its-upcoming-super-app-170000019.html?src=rss

Google Chrome makes it easier to wrangle different tabs in AI Mode

Love 'em or hate 'em, no modern browser is complete without robust tab support, and so too would it seem Google's AI Mode. Starting today, the company is rolling out an update to users in the US that makes the tool better at interacting and understanding tabs. 

To start, the next time you use AI Mode on Chrome for desktop and click on a link, the chatbot will open a new side-by-side interface that allows you to both browse the new webpage and ask questions of AI Mode. The connection allows the chatbot to maintain the context of the search that brought you to that website in the first place. 

For instance, say you're looking for a new coffee maker to buy for your apartment. After AI Mode finds a handful of different models for you to compare, you can click on one to go to the manufacturer's website and ask additional questions of the chatbot like "how easy is this to clean?" Thanks to the expanded context window, you don't need to refer to the specific name of the model.   

Meanwhile, if you have an existing tab or group of tabs that you'd like AI Mode to factor into a new search, you can do that now too. From the redesigned Plus menu, just click the new option that's there. While you're in the Plus menu, you can also prompt AI Mode to consider other materials, including images and PDFs, alongside any relevant tabs.   

In testing, Google says users found the integration translated to less tab switching, and made it easier to focus. Mike Torres, vice-president of product for Chrome, said the new features represent a broader effort by Google to bring practical AI capabilities to its web browser. Torres added the company would soon bring today's updates to more places around the world.

This article originally appeared on Engadget at https://www.engadget.com/ai/google-chrome-makes-it-easier-to-wrangle-different-tabs-in-ai-mode-170000914.html?src=rss

Canva starts previewing a more powerful version of its AI assistant

Adobe isn't the only company releasing a new AI assistant this week. Ahead of its Create event in Los Angeles today, Canva announced Canva AI 2.0. Building on its existing AI assistant, the company is billing the release as its most significant update since the platform first launched in 2013, and the culmination of years of investment to build its own foundational design models. 

As you might imagine, it all starts with a conversational interface that allows you to describe an idea or goal and the system will start generating a design to match. Under the hood, there's a new orchestration layer that allows the model to use all of Canva's disparate tools to accomplish complex, multi-step tasks. For instance, the company suggests you could use Canva AI to create a multi-channel advertising campaign, and the software will generate everything you need to get that off the ground. 

For brands, Canva AI 2.0 can adapt to their design needs.
For brands, Canva AI 2.0 can adapt to their design needs.
Canva

If edits are required, the company says Canva AI avoids one of the pitfalls of many other image generation models. It's possible to edit every visual element the system generates, just like if they were created with a traditional image editor. As a result, you can do things like swap out images and tweak fonts without affecting any other part of a design. To bring everything together, Canva has built persistent memory into the tool. The more you use Canva AI, the better the system will get at applying your personal taste and style to future generations. According to the company, it also has a context window that is long enough to maintain coherence until you arrive at a final design.    

Alongside those enhancements, Canva is adding support for new workflows that expand what you can do with its software, starting with connections that allow its models to pull data from other apps, including Notion, Slack, Zoom, Gmail, Google Calendar and more. Users can also schedule tasks for Canva AI to complete in the background, and the company has even baked in deep research capabilities into the tool.

The coding function Canva previously offered has been upgraded to include support for HTML imports, allowing users to bring any HTML file or AI-generated experience into Canva's visual editor to tweak the design of it without breaking things. For brands, the company is also offering a tool that can process their visual identity and apply it to new and existing designs.   

Canva's updated coding agent now support HTML imports.
Canva

As a casual observer, it might seem like Canva is trend chasing, but Danny Wu, the company's head of AI, argues the new AI tools represent a natural evolution for Canva. "This is something we've been dreaming of and working towards for quite a while," he tells Engadget. "Even before ChatGPT was a thing, we were thinking, 'what if we don't have a template that matches your needs?' … So I wouldn't describe this as a pivot or shift, we've been wanting to offer these kinds of capabilities all along as part of our mission to make design simple."

If you want to give Canva's new tools a try for yourself, Canva AI 2.0 is available as a research preview starting today. The first 1 million people who visit the Canva website will get first access, with availability gradually expanding to more users over the coming weeks. As before, access to Canva’s AI features remains included in the company’s free offering, though it’s also introducing a new AI Pass add-on that significantly increases rate limits for users.

This article originally appeared on Engadget at https://www.engadget.com/ai/canva-starts-previewing-a-more-powerful-version-of-its-ai-assistant-130000966.html?src=rss