Anthropic's draft AI bill sets 50-death, $1 billion bar to halt models

The bar: Under the draft, the government could block a model's release or use only if it poses a foreseeable, imminent risk of more than $1 billion in damages, or of killing or seriously injuring more than 50 people, according to Politico's report as summarized by Reuters. The risk would also have to come from one of three sources: the creation or release of a biological weapon, a model evading its developer's controls, or a cyberattack.

Who could pull the plug: The Commerce Department could ask a federal judge in the U.S. District Court for the District of Columbia to halt a model, and that judge could order a suspension. The draft explicitly bars any other federal department, agency or officer from suspending a model's development, deployment or internal use, Reuters reported. The Commerce Department did not respond to Reuters' request for comment.

Testing and reporting: Independent evaluators registered with Commerce would periodically review the labs and their frontier models. Labs could pay those evaluators as long as the payment isn't tied to the outcome, and evaluators who find an "imminent enumerated risk" would have to tell Commerce within 72 hours. An Anthropic spokesperson told AFP the bill "grants the federal government the ability to block or deter dangerous deployments of the most powerful models" and requires developers to test them, publish their findings, submit to outside evaluation and keep a strong security program.

States: The draft would override some state AI rules on risk disclosure, independent evaluation and incident reporting, including rules in force in California, AFP reported.

Where it stands: It is a company proposal, not an introduced bill, and Congress is out of session until after the Nov. 3 midterm elections, AFP noted. It joins other plans: the House FRONTIER Act from Reps. Lori Trahan (D-Mass.) and Jay Obernolte (R-Calif.), which would authorize government action against risky models and override some state laws, and a Senate effort on catastrophic AI risks involving Majority Leader John Thune (R-S.D.) and Sens. Amy Klobuchar (D-Minn.) and Ted Cruz (R-Texas), Interesting Engineering reported, citing Politico.

Why now: Anthropic was forced by Washington to suspend its most advanced models for nearly three weeks in June, AFP reported, and the draft would close off that kind of action by any agency other than Commerce acting through a court. The company is also preparing a November IPO at a valuation that could approach $2 trillion, Business Insider reported. Nothing in the reporting ties the draft's timing to the listing.

The strongest case from each side: Supporters say a court-only path with narrow, defined triggers gives the industry predictable rules, prevents sudden executive action like the June shutdown, and replaces a patchwork of state laws with one standard. Charlie Bullock of the Institute for Law & AI called the draft "pretty reasonable" and "more than other frontier labs have done," AFP reported. Critics say the 50-person and $1 billion thresholds are so high that the government could rarely act in time, that lab-paid evaluators invite conflicts, and that the drafter is the company being regulated. Bullock also said on X that the draft is "weaker" than the FRONTIER Act, according to AFP.

What the evidence shows: Anthropic wrote in June 2026 that "absent a strong federal law, states should retain full authority to legislate," AFP reported. The draft is consistent with that only if lawmakers judge it to be a strong federal law, and that is the point critics dispute.

What's still unknown: Which lawmakers received the draft and whether any will introduce it; how the Commerce Department and the White House view it; and whether rival labs such as OpenAI and Google back the same thresholds.

This story rests on Politico's report of a confidential draft, which cited three people with knowledge of Anthropic's outreach, and on Anthropic's confirmation to Reuters and AFP that the draft is circulating. Details of the bill's text come from Politico's reporting as summarized by Reuters and AFP; US Brief has not seen the draft. It has not been introduced in Congress and could change. Anthropic has not said which lawmakers received it.