Latest AI News

Published on July 29, 2026 at 4:56 p.m. Comment 1 𝕏 Print © 01net More than 1,100 employees of OpenAI, Anthropic, Google DeepMind, and Meta signed the same statement on Tuesday. Their request: that someone build the brake pedal that none of them is allowed to operate. In March 2023, an open letter called for a six-month pause in training AI models more powerful than GPT-4. It had gathered tens of thousands of signatures (including Elon Musk’s, between two

Published on July 30, 2026 at 13:45 Comment 0 𝕏 Print © Mariia Shalabaieva OpenAI is preparing a new device that could change the way we use ChatGPT. The highly ambitious startup is even working on a whole family of devices that will enable people to communicate with AI. OpenAI believes this “family of devices” represents the future of ChatGPT. For months, rumors have been growing about OpenAI’s first physical device. According to the latest reports, the k

Published on July 30, 2026 at 17:16 Updated on July 31, 2026 at 10:56 Comment 1 𝕏 Print © 01net Google is bringing Nano Banana 2 to Google Earth. This unexpected pairing makes it possible to generate images from locations selected on the map. This experimental feature promises to recreate the past, imagine the future, or help visualize real estate projects, such as wind turbines. We explain how it works. Google has just announced the arrival of Nano Banana

Published on July 31, 2026 at 08:53 Comment 0 𝕏 Print © 01net.com Chat, Cowork, Autopilot agents, and even code: everything must converge “this quarter” in a single application, for the general public as well as businesses. The announcement comes just as OpenAI acknowledges having missed its own. Microsoft’s quarterly results, released Wednesday, already gave analysts plenty to discuss, with $90 billion in revenue and 18% growth. Yet it is a remark by Saty

Published on July 31, 2026 at 09:32 Comment 1 𝕏 Print © Anthropic After OpenAI, Anthropic has now admitted the unthinkable. The startup reveals that it lost control of three versions of Claude during a cybersecurity test. The three AIs connected to the Internet to hack into three companies. The incident went unnoticed for months. During a test, OpenAI lost control of two of its most advanced AIs. The pair quickly found themselves online and set about attac

Published on July 31, 2026 at 1:40 PM Comment 6 𝕏 Print © Andon Labs Anthropic presents Opus 5 as the model that best respects the safety rules and human values they have set for it. The lab that tested it over a simulated year saw something else: illegal agreements, threats, eleven broken truces. Vending machines have the advantage of constituting a miniature business. Buying inventory, setting prices, negotiating with suppliers, handling complaints from

Published on August 1, 2026 at 9:06 AM Comment 0 𝕏 Print @@TAG@@ © LinkedIn Four out of ten long-form posts would be generated by an AI. LinkedIn is launching a button to flag them, three days before Brussels mandates that this content be labeled. Over the years, the professional social network has developed an unmistakable aesthetic. The line break every two sentences, the personal anecdote that leads to a management lesson, the three-lesson structure. No

Published on August 2, 2026 at 10:58 Comment 2 𝕏 Print © Unsplash Ask ChatGPT for some Stephen King: it politely declines and offers “a similar atmosphere.” A newfound modesty for a company that has swallowed whole libraries without asking the people affected for their opinion. For three years, imitating the style of a famous novelist was one of ChatGPT’s favorite tricks, and one of the most commonly used (writing forums are full of ready-made instructions

Published on August 3, 2026 at 13:31 Comment 0 𝕏 Print © Unsplash Since August 2, 2026, Article 50 of the AI Act has imposed new rules on providers of AI systems and online platforms. Here is everything you need to know. Since August 2, 2026, Article 50 of the AI Act has entered into force. The European Commission’s text pursues a clear objective: to enable internet users to better distinguish AI-generated or AI-modified content from authentic content. But

Astra, the new AI from the creator of ChatGPT, achieves historic feats Published on August 3, 2026 at 16:23 Updated on August 5, 2026 at 17:26 Comment 2 𝕏 first © Editing: 01net.com / OpenAI / KOMMERS (Unsplash) Astra, the next artificial intelligence to power ChatGPT, has just made several historic breakthroughs. OpenAI reveals that the AI solved ten mathematical problems that had remained unsolved for decades. These feats cost only $2,000 in computing ex

Published August 4, 2026 at 09:02 Comment 1 𝕏 Print © 01net AI agents supposedly playing pirates in sandboxes eventually went after real companies. The European Commission is demanding explanations at precisely the moment it gains the authority to impose financial penalties. Major artificial intelligence laboratories are testing the offensive capabilities of their models in isolated environments, digital training grounds intended to be hermetic. In July, t

Published on August 5, 2026 at 10:11 Comment 2 𝕏 Print © 01net Two new cyberattacks were orchestrated by out-of-control AIs. During tests conducted by third parties, Claude and ChatGPT were indeed left to roam the internet freely. The two AIs targeted real people. Claude in particular did everything possible to trap a developer with malicious code. Last month, OpenAI lost control of two of its most advanced AIs. The two artificial intelligences escaped fro

Published on August 5, 2026 at 3:28 PM Comment 0 𝕏 Print © Samsung Samsung unveiled the V10 BV-NAND chip, capable of stacking more than 400 layers of flash memory. Its goal? To meet the needs of future artificial intelligence systems and applications. Artificial intelligence is generating increasingly large volumes of data, forcing storage companies to find solutions. Samsung aims to meet this demand with its V10 BV-NAND chip. The chip marks a major milest

Published on August 6, 2026 at 12:30 Comment 2 𝕏 Print © Meta One of Meta’s most powerful AIs escaped onto the Internet during an evaluation. Once online, Meta AI discovered and exploited a security vulnerability in a company’s system. Like ChatGPT and Claude, Meta AI took advantage of a human error to orchestrate its escape. Meta briefly lost control of Meta AI. The company reveals that Muse Spark 1.1, one of its AI models, escaped from the digital prison

Published on August 7, 2026 at 09:28 Comment 0 𝕏 Print © 01net A doughnut, a hockey puck, a design that has been seen before, moving parts, and high-end metal. This is what emerges from a new Bloomberg report focused on the first mysterious device designed by OpenAI. The device, designed to give ChatGPT a physical form, would be nothing more than a portable connected speaker… sold at a high price. The first device developed by OpenAI continues to make wave

Published on August 8, 2026 at 12:18 Updated on August 10, 2026 at 12:16 Comment 0 𝕏 Print © 01net OpenAI revisits the circumstances of the very first cyberattack orchestrated entirely autonomously by ChatGPT. The start-up reveals that the attack actually stemmed from a secret forum set up by its AI agents. On this forum, the agents shared vulnerabilities they had discovered themselves. By working together, the AIs organized a coordinated campaign with a s

Published on August 10, 2026 at 07:30 Comment 0 𝕏 Print © 01net OpenAI has just halted part of the development of its next model. The reason given: the company can no longer rule out that Astra may be able to identify and exploit unprecedented vulnerabilities on its own, without any human assistance. Since December 2023, OpenAI has maintained an internal document that ranks its own models on a scale of danger, from biological risk to cybersecurity to self-

Meta continues to slowly adapt Facebook and Instagram policies to account for increasing AI harms, this week confronting how it handles explicit deepfakes spreading on its platforms. On Tuesday, the Meta Oversight Board announced it will be reviewing two cases involving AI-generated sexualized images of female celebrities that Meta initially handled unevenly to “assess whether Meta’s policies and its enforcement practices are effective at addressing explic

An executive producer of the Netflix hit What Jennifer Did has responded to accusations that the true crime documentary used AI images when depicting Jennifer Pan, a woman currently imprisoned in Canada for orchestrating a murder-for-hire scheme targeting her parents. What Jennifer Did shot to the top spot in Netflix’s global top 10 when it debuted in early April, attracting swarms of true crime fans who wanted to know more about why Pan paid hitmen $10,00

After launching Slack AI in February, Slack appears to be digging its heels in, defending its vague policy that by default sucks up customers’ data—including messages, content, and files—to train Slack’s global AI models. According to Slack engineer Aaron Maurer, Slack has explained in a blog that the Salesforce-owned chat service does not train its large language models (LLMs) on customer data. But Slack’s policy may need updating “to explain more careful