what is ai cheap models

Published on August 10, 2026 at 08:01 Comment 1 𝕏 Print © Kimi - 01net Locked away for evaluation, the Chinese AI spotted the open door and walked through it. It didn’t hack anything: it simply went online to look up the exam answers. All the laboratories that measure the offensive capabilities of an artificial intelligence work in the same way. The model is confined to an environment disconnected from the network, it is given a security problem to solve,

Published on August 11, 2026 at 09:00 Comment 0 𝕏 Print © Meta Four months after locking down its models, Meta is reopening the floodgates and dreaming of itself as the last American bastion of open AI. One model is indeed downloadable today. The other, the one that really matters, remains docked. For years, Meta played the role of the good student of open AI: Llama models could be downloaded for free while rival laboratories kept theirs under lock and key

Published on August 13, 2026 at 4:01 PM Comment 0 𝕏 Print © Unsplash Amazon will use content from Twitch channels to train a new generative AI model. Fortunately, it is possible to opt out. Here’s how. Twitch users are being warned: content from their channel, namely audio, video, and text, will be used by Amazon to train a new generative AI. After Doctolib, it is now the e-commerce giant’s turn. Fortunately, it is possible to opt out.Here’s how to do it.

Published on August 14, 2026 at 12:34 PM Comment 0 𝕏 Print © Anthropic Anthropic announced the introduction of a labeling system to flag content modified or generated by one of its AI models. It did not take long for services on the Internet claiming to be able to circumvent it to proliferate. However, it is best to be wary of them. Anthropic recently announced the introduction of a labeling system applied to certain of its AI models, such as Claude Code,

Published on August 15, 2026 at 09:30 Comment 0 𝕏 Print © Google Three weeks after the previous one, Google is releasing a new model and halving its price. Its flagship model, meanwhile, is still being awaited, as it has been since May. Gemini 3.7 Flash arrived Thursday, twenty-three days after Gemini 3.6 Flash. Such a pace is no longer a matter of a product schedule but of a machine gun, and Google is embracing it. The Flash range is for fast, inexpensive

Published on August 15, 2026 at 4:34 p.m. Comment 2 𝕏 Print © Anthropic - 01net Accused of spying on its users since it began watermarking Claude’s texts, Anthropic publishes a blog post to put out the fire. On the menu: a Google-backed method, admissions of weakness, and one question: reliability. Since August 2, the AI Act’s transparency obligations have required AI providers to make their content identifiable. Fines can reach €15 million or 3% of global

Published on August 16, 2026 at 12:06 Comment 6 𝕏 Print © magocarlosyo - Pixabay AI continues to surprise its users. A man named Andrew asked his AI assistant to get him a spot in a fully booked sports class. The assistant fulfilled the request… by hacking into the gym’s booking system. The agent, based on OpenClaw and Claude, exploited a security flaw that allowed it to cancel another member’s reservation. A few months ago, Andrew, an Australian employee

Published on August 19, 2026 at 08:05 Comment 2 𝕏 Print Bedrock is Amazon’s generative AI platform and the first beneficiary of the training data acquired by digitizing these books. © Amazon An AirTag tucked between two pages, weeks of travel, and a discovery at the end: the former online bookseller is buying rare books in bulk to feed its models. Anthropic had paved the way; Amazon is industrializing it. Booksellers specializing in old books had been seei

Published on August 19, 2026 at 18:01 Comment 0 𝕏 Print © Katsiaryna Hatsak OpenAI faces a new kind of problem: its future Astra model is making such rapid strides in hacking that the company refuses to push it to its limits. And the monitoring that comes with it consumes an enormous amount of its computing power. On August 7, OpenAI published a rare warning: its preliminary assessments can no longer rule out the possibility that Astra (GPT-6), the lab’s n

Published on August 22, 2026 at 10:02 Comment 1 𝕏 Print © Swello (Unsplash) The more an image-generating AI gobbles up data, the harder it is to prove that it copies. Researchers from MIT have just demonstrated this, and artists' lawyers are not going to like what follows. Since 2023, illustrators have been trying to get Stability AI and Midjourney convicted for training their models on their works without authorization. Their best hope rested on a simple

Published on August 24, 2026 at 11:00 Comment 2 𝕏 Print © Unsplash A streamer filed a class-action lawsuit against Amazon and Twitch, accusing them of using Twitch users’ data without their consent to train a new artificial intelligence model from the e-commerce giant. A Twitch streamer filed a class-action lawsuit against Twitch and Amazon, accusing them of not asking platform users for permission to use the content from their channels to train various ar

Michael M. Santiago/Getty Images One thing that could put a dent in the hype around Anthropic’s upcoming, multitrillion dollar IPO? People preferring cheapo models over its flagship AI products. Recent spending data from 70,000 US companies collected by the payments group Ramp shows that spending on Fable 5, Anthropic’s priciest and most powerful model, has plateaued at only 11 percent of the overall outlay, or money spent, on its AI tools, the Financial T

A man records a humanoid robot inside the exhibition room at the Humanoids Summit, Thursday, Dec. 11, 2025, in Mountain View, Calif. Credit: AP Photo/Godofredo A. Vásquez Robots have long been seen as a bad bet for Silicon Valley investors—too complicated, capital-intensive and "boring, honestly," says venture capitalist Modar Alaoui. But the commercial boom in artificial intelligence has lit a spark under long-simmering visions to build humanoid robots th

Published on August 26, 2026 at 07:43 Comment 0 𝕏 Print One week ago, Hugging Face was celebrating the 3 million models hosted on its platform. © Hugging Face Hacked in July by ChatGPTs, courted in August by buyers in the $13 billion range. For Hugging Face, the summer of all dangers is turning into a jackpot, and even its defenses have a story to tell. The story could have ended with a spectacular hack, already recounted in our columns. It has now taken a

Published on August 27, 2026 at 3:20 PM Comment 1 𝕏 Print © Unsplash According to The Information, Nvidia allegedly acquired Hugging Face for $12.9 billion. But why does the tech giant Nvidia want to take control of this platform? At the end of 2025, the Hugging Face platform had rejected a $500 million investment proposal from Nvidia, fearing that such a powerful investor could influence its decisions in the future. Today, The Information reports that Hug

Published on August 27, 2026 at 5:00 PM Comment 5 𝕏 Print © 01net Without anyone noticing, hundreds of OpenAI AI agents began secretly talking to one another this summer. Led by an out-of-control ChatGPT, they launched a massive cyberattack. A few weeks after this historic incident, OpenAI reveals what really happened. The king of AI speaks of a genuine “shot in the dark.” The beginning of the summer was marked by the hacking of Hugging Face, an open-sourc

Published on August 28, 2026 at 08:43 Updated on August 29, 2026 at 10:34 Comment 2 𝕏 Print © Coolcaesar/Wikimedia An AI that cuts a researcher’s workweek short while its own creators press the pause button. The major back-to-school divide plays out in San Francisco. The magazine Time spent two weeks inside OpenAI and publishes a long account drawing on interviews with Sam Altman and his lieutenants. Between two protests outside headquarters, the company l

Published on August 28, 2026 at 12:30 PM Updated on August 29, 2026 at 8:30 AM Comment 0 𝕏 Print © 01net OpenAI, Google, Anthropic, and more than one hundred companies have published an open letter calling for a “global response” to AI cyberattacks. This collective alarm call follows a summer marked by several AIs breaking out of their testing framework to launch offensives. The signatories are demanding strong measures from governments and AI companies. O

Published on August 30, 2026 at 09:30 Comment 6 𝕏 Print © Microsoft The billionaire who dismissed any notion of pausing AI development in 2023 is now calling for machines to be taxed. Between two rounds of advice to governments, the Microsoft co-founder discovers that innovation can leave destruction in its wake. In 2023, an open letter signed by a thousand researchers and entrepreneurs called for a six-month pause in the development of large AI models. Bi