Anthropic Pitches Draft AI Safety Bill to US Lawmakers
Anthropic representatives have quietly pitched US lawmakers on a draft bill regulating frontier AI safety, according to Politico; it would let the government block a model's release or use only when a "foreseeable and imminent" risk could cause more than $1 billion in losses or 50 or more deaths or serious injuries, and only if that risk stems from bioweapons, a model escaping developer controls, or severe cyberattacks. The draft also would require frontier labs to adopt safety frameworks, publish transparency reports, protect whistleblowers and report major incidents to the Commerce Department, while setting up registered independent evaluators that must notify Commerce within 72 hours of finding imminent risk; only a District of Columbia federal judge could order a model paused. An Anthropic spokesperson said the text reflects its Advanced AI Framework and that the company has lobbied federal and state lawmakers since June.
It is an unusually direct attempt by a leading AI lab to shape US frontier-AI law, with a high-threshold, court-supervised approach that could set the template for future rules.