Anthropic is no longer only commenting on how Congress should regulate frontier AI; it has drafted language of its own. Company representatives have been showing lawmakers on Capitol Hill a draft bill on frontier AI safety and security, Politico reported on Friday, October 16, citing three people with knowledge of the effort. Reuters followed with a statement from an Anthropic spokesperson, who said the bill represents proposals the company set out in its Advanced AI Framework earlier this year.
As described by Politico, the draft sets a high bar for government intervention. Authorities could block the release or use of a model only if it posed a foreseeable and imminent risk of causing more than $1 billion in damage or of killing or seriously injuring more than 50 people. That risk would have to come from one of three sources: a model creating or releasing a biological weapon, a model escaping its developer's control, or a cyberattack.
Oversight would run through independent evaluators registered with the Commerce Department. They would periodically assess AI developers and their frontier models, and labs could pay them as long as payment did not depend on the outcome of the evaluation. An evaluator that found an "imminent enumerated risk" would have to notify Commerce within 72 hours of finishing its assessment. Commerce could then ask a federal judge in the District of Columbia to suspend a model, and the draft would explicitly bar any other federal department, agency or official from halting a model's development, deployment or internal use.
Anthropic's spokesperson framed the proposal as giving the government the ability to block or deter dangerous deployments of the most powerful models, while requiring developers to test them, be transparent about results and submit to independent evaluation. Those elements match the company's published framework from June 2026, which calls for catastrophic-risk testing of the most capable models, engagement of qualified external evaluators, ongoing disclosure of results and incidents, and accountability to a government agency.
The reported draft is narrower than that framework in at least one respect. The framework names four risk categories for developer obligations, including automated AI research and development alongside biological weapons, offensive cyber operations and loss of control. Politico's account of the bill's intervention trigger lists only three. The framework also argues that Congress should not preempt state AI laws unless it enacts a federal regime at least as rigorous as the one Anthropic proposes; the reports do not say how the draft handles preemption.
The pitch lands in a crowded Senate debate. In July, Punchbowl News reported that a frontier AI bill being negotiated by Senate Majority Leader John Thune, Commerce Committee Chair Ted Cruz and Senator Amy Klobuchar stalled partly over disagreements with Anthropic, which, according to that report, objected to giving the Commerce Department an ongoing injunction power. Anthropic's own draft, as described, routes any suspension through a court order requested by Commerce rather than direct agency action.
Much remains unknown. The text has not been published, no sponsor or co-sponsor has been named, and it is unclear which lawmakers received it or how they responded. The Commerce Department did not respond to requests for comment, according to Reuters. The detailed thresholds and procedures come from Politico's sources rather than from a document Anthropic has released, so they could change before any bill is introduced.