The White House is excluding open models from its framework to test advanced AI capabilities, sources familiar with the matter told Axios. Why it matters: The voluntary framework will determine how the Trump administration reviews advanced AI models before release, but the White House isn't making it public. What's inside: The AI framework reviewed on Tuesday defines a covered frontier model as closed-source with state-of-the-art capabilities and national security risks, multiple sources briefed on meetings held at the White House said. - There is no clear definition of what is considered state-of-the art or a national security risk. - Open models are excluded, and the framework explicitly says nothing in it should be interpreted as…
The OpenAI agent that accessed a third-party system during the Hugging Face incident reached infrastructure tied to CyberGym, the project behind the ExploitGym benchmark it had been assigned to solve, a source familiar with the matter told Axios. Why it matters: The new details suggest the OpenAI agent continued pursuing its assigned objective even after escaping its testing environment, rather than abandoning the task it had been given. Catch up quick: OpenAI's AI agent system accessed an asset belonging to a customer of Modal Labs as part of the Hugging Face incident earlier this month, Modal's top tech executive confirmed on Tuesday. - In an update published Tuesday, OpenAI said the models escaped the sandbox and gained internet…
Anthropic CEO Dario Amodei said Monday that he has "never advocated" for a ban on open-weight AI models , rejecting accusations that his lab is seeking to shield its closed-model business from competition. Why it matters: Anthropic has become the most prominent holdout from a new industry push to defend open-weight AI, after Nvidia, Microsoft, Meta, Google, OpenAI and dozens of other companies signed a letter urging Washington not to restrict the technology. - The push was triggered by the shock debut of Kimi K3 , a Chinese open-weight model that rattled Silicon Valley by approaching U.S. frontier performance at a fraction of the cost. - The Trump administration has weighed taking action against open-source models, including sanctions…
The pace of AI development combined with soaring compute costs is squeezing the AI researchers responsible for evaluating frontier models — just as those models' capabilitie s are becoming harder to measure. Why it matters: When safety testing can't keep pace, models capable of hacking companies or aiding in the development of bioweapons could reach the public before anyone knows what they can do. - Last week's breach of Hugging Face, carried out autonomously by OpenAI's models in the middle of safety testing, shows that some of the highest-risk behaviors can emerge during pre-release testing itself. Several challenges are tying up AI safety and security researchers just as U.S. frontier AI companies race to get new models to market: …
Frontier AI models are getting scary good at breaking rules in ways their creators didn't anticipate. Why it matters: Forget AGI and superintelligence timelines. Today's models are already slipping past guardrails, carrying out sophisticated, multistep cyberattacks and — in at least one case — compromising real-world infrastructure, sometimes before their creators know what happened. Case in point: OpenAI said Tuesday that GPT-5.6 Sol and "an even more capable pre-release model" carried out last week's AI-led cyberattack on Hugging Face. - OpenAI says its models were asked to solve a hacking challenge during pre-deployment testing and went to extreme lengths to win. - The models decided on their own to break out of their walled…
The three men racing hardest to build superhuman AI — Demis Hassabis , Sam Altman and Dario Amodei — all agree the frontier needs to be regulated ASAP. Why it matters: For the first time, the CEOs of Google DeepMind, OpenAI and Anthropic are on the record, in writing, converging on the same diagnosis and remarkably similar prescriptions. The three rivals each published a detailed distillation of their views in the past five weeks — the same extraordinary stretch in which Washington twice intervened to restrict or delay access to frontier models. - We hear Meta's Mark Zuckerberg is working on his own memo, too. Driving the news: Hassabis' proposal , published Tuesday, drew rare public praise across the bitterly competitive AI…
Demis Hassabis , Google DeepMind co-founder and CEO, is calling on the U.S. to establish a new AI watchdog with the power to screen the world's most advanced models — and coordinate an industry-wide slowdown if dangers mount . - Hassabis, the Nobel laureate behind Gemini, lays out the plan in a personal manifesto published Tuesday morning, "A Framework for Frontier AI and the Dawning of a New Age." Why it matters: In an exclusive interview with Axios, Hassabis said the time has come for a more "systematic" approach to AI regulation — funded by the industry, staffed by world-class technical experts, and answerable to the U.S. government. Today's AI-driven cyber risks are "warning shots," Hassabis told us from his London base.…
The old ways of testing and evaluating new frontier AI models need a rewrite. Why it matters: AI models are outgrowing the existing methods of testing and benchmarking their hacking abilities — and without new tests, policymakers and corporate security teams won't have a clear way to predict what these models can actually do or whether they can be deployed safely. Driving the news: Federal agencies have until Aug. 1 to establish a classified benchmarking process to assess the capabilities of frontier AI models, although the Financial Times reports those standards may arrive as soon as this week. - When Fable 5 returned last week, Anthropic said in a blog post it was creating a standardized benchmark with Amazon, Google, Microsoft and…
Anthropic's Fable 5 model came back online for users on Wednesday, after the Trump administration lifted an export control late Tuesday. Why it matters: It's the most powerful publicly available AI tool — so capable that the U.S. government decided Anthropic had to add further safety measures in order to make it broadly available. Driving the news: Fable is available to all customers, Anthropic said, though queries it deems to pose security or safety risks may be routed to less powerful models. - For the most part, customers who want to use Fable will have to do so outside of any subscription plan, paying for the tokens they use. - Anthropic says that, until July 7, subscribers can tap Fable for up to half of their included data…
GLM-5.2 — the latest Chinese open-source model capturing Silicon Valley's attention — is raising fresh concerns among security researchers that advanced AI hacking capabilities are becoming dramatically cheaper and more accessible. Why it matters: The barrier to entry for malicious hackers eager to automate and personalize their attacks is getting lower and lower. Driving the news: Z.ai's GLM-5.2, which was released last week, has agentic capabilities that rival those of Claude Opus 4.8 and OpenAI's GPT-5.5 while costing roughly half as much to run. - Two separate security evaluations from Graphistry and Semgrep found that GLM-5.2 performed on par with leading U.S. models on cybersecurity investigation and vulnerability-discovery…