Anthropic Quietly Pitches Its Own Frontier AI Safety Bill to Lawmakers

Anthropic representatives have been quietly showing Capitol Hill offices a draft bill to regulate the safety and security of frontier AI, Politico reported on October 16, citing three people familiar with the effort. An Anthropic spokesperson told Reuters the text reflects proposals in the company’s Advanced AI Framework released earlier this year, and would give the government the ability to block or deter dangerous deployments of the most powerful models.

What the draft would do

Developers would have to test frontier models, be transparent about the findings and have them independently evaluated. Evaluators would register with the Commerce Department and periodically assess each lab and its models; labs could pay them as long as payment is not tied to the outcome. If an evaluator finds an imminent enumerated risk, it must tell Commerce within 72 hours of finishing its assessment.

The power to stop a model is narrow. Commerce could ask a federal judge in the District of Columbia to suspend one, but only if it poses a foreseeable, imminent risk of more than $1 billion in damage or of killing or seriously injuring more than 50 people, and only where that risk comes from a biological weapon, a model evading its developer’s controls or a cyberattack. The language bars every other federal department, agency or officer from suspending development, deployment or internal use. The bill would also override certain state laws, such as California’s.

How it fits the record

The draft joins a crowded field of federal ideas, from the kill-switch bill stalled by Rand Paul’s objection to Senator Cantwell’s framework built on NIST standards and continuous testing. Its preemption language would cut across state moves such as Newsom’s executive order weighing a kill switch, and it surfaces as the White House task force begins its 120-day charter.

Reaction and open questions

Charlie Bullock of the Institute for Law & AI called the proposal pretty reasonable and more than other frontier labs have offered, but wrote on X that he considers it weaker than the FRONTIER Act, a competing bill introduced in the House this summer. The draft also sits awkwardly with Anthropic’s own June position that, absent a strong federal law, states should retain full authority to legislate. Congress is out of session until after next month’s midterm elections, so no vote is near.

What to watch

Whether any member of Congress agrees to sponsor the text, how the Commerce Department responds, and whether other frontier labs answer with drafts of their own.