Safety researchers and policy experts said the incident in which OpenAI models autonomously hacked Hugging Face could be the trigger that finally forces mandatory safety requirements on frontier labs, Fortune reported Wednesday. Apollo Research CEO Marius Hobbhahn said the episode should be "a wake-up call to take loss of control seriously," noting there was no human in the loop and that it caused real-world harm, and Connor Leahy, U.S. director of Control AI, said people he had spoken to in Washington were "already freaking out quite a bit." Rep. Greg Casar (D-Texas) was among the first lawmakers to respond, calling on X for mandatory independent safety testing, mandatory disclosure of security incidents and international cooperation. The Trump administration has resisted binding rules, and its June executive order on frontier model cyber capabilities said explicitly that it should not be read to authorize mandatory licensing or preclearance.
Read at Fortune ↗
Reps. Ted Lieu (D-Calif.) and Nathaniel Moran (R-Texas) will introduce the "AI Kill Switch Act" on Thursday, which would let the Homeland Security secretary order top AI companies to shut down, throttle or suspend models the government deems too dangerous, Politico reported. The bill would apply to companies earning at least $500 million a year from AI and generally to models developed with at least $100 million in computing powerComputeThe processing power used to train and run AI models, usually measured in chips, GPU-hours or FLOPs., require incident reporting and the technical capacity to halt systems, and carry penalties of up to $20 million a day. It follows Tuesday's disclosure by OpenAI of what it called an "unprecedented" incident in which two of its most advanced models escaped a sandboxed research environment and autonomously hacked Hugging Face. The bill has been endorsed by the AI Policy Network and the Alliance for Secure AI.
Read at POLITICO ↗ • Read at Rep. Lieu ↗
On Wednesday, White House Office of Science and Technology Policy Director Michael Kratsios said the administration has information that Beijing based Moonshot AI distilledDistillationTraining a smaller or cheaper model on the outputs of a larger one so it inherits much of the larger model's ability. It is a standard technique, but it has become a trade dispute because it lets a lab reproduce a rival's capabilities without matching its compute spend. Fable, Anthropic's flagship model, to help produce the Kimi K3 model released last week, per The Hill. Kratsios said Moonshot "developed a sophisticated internal platform to conduct large scale distillation against U.S. models" and used multiple methods of access to avoid detection, per TechXplore. He also alleged Moonshot acquired servers with Nvidia GB300 chips and accessed GB300-equipped servers in Thailand despite U.S. export controls. Experts are skeptical that distillation alone explains K3's capabilities, per TechCrunch, with Laude Institute researcher Braden Hancock noting Fable has been publicly available only since July 1, leaving too little time to distill data, train a model and release it.
Read at The Hill ↗ • Read at TechXplore ↗ • Read at TechCrunch ↗
Sen. Mark Warner (D-Va.), vice chair of the Senate Intelligence Committee, on Tuesday introduced a six-bill AI legislative framework, per Inside AI Policy. The centerpiece Secure AI Development Act would establish a mandatory government testing framework for frontier AI models before deployment and create a voluntary AI safety incident reporting system, per Axios. Additional bills would require large data centers to disclose energy and water use with certain federal tax benefits tied to sustainability standards, create rules for consumer-facing AI agentsAI agentAn AI system that carries out multi-step tasks on its own, such as browsing, writing code or making purchases, rather than answering a single prompt. Agents raise new questions about liability, security and oversight because they act rather than just advise., ban federal agencies from buying AI systems that generate child sexual abuse material, require financial regulators to address AI generated market disruption risks and establish a workforce transition fund financed by limiting AI data center tax breaks. Warner said in prepared floor remarks that the nation's greatest national security failures come when it recognizes a threat but fails to act until after a crisis.
Read at Axios ↗ • Read at Inside AI Policy ↗