Anthropic has presented US lawmakers with a draft bill proposing new rules for frontier artificial intelligence models, according to a report by Politico cited by Reuters on 16 October 2026.
The proposal would allow the US government to block the release or use of an AI model only when it poses a foreseeable and imminent risk of causing more than $1 billion in damage, or killing or seriously injuring more than 50 people.
The reported draft outlines a potential framework for government intervention as policymakers assess the risks associated with increasingly capable AI systems.
Conditions for Government Intervention
According to the Politico report, the proposed restrictions would apply to risks arising from the creation or release of a biological weapon, an AI model evading its developer’s controls, or a cyberattack.
The proposal would also require developers of frontier AI models to test their systems, disclose findings and undergo independent evaluations. The aim would be to establish greater accountability for the development and deployment of powerful AI systems.
Independent evaluators would register with the US Department of Commerce and periodically assess AI developers and their frontier models. Companies could pay for these evaluations, provided that payment was not linked to the outcome.
If an evaluator identified an imminent risk covered by the proposal, it would have to report the findings to the Department of Commerce within 72 hours of completing its assessment.
Limits on Regulatory Powers
The draft would give the Department of Commerce the ability to ask a federal judge to suspend a model after determining that it presents the specified risk.
However, the reported language would prevent other federal departments, agencies and government officers from independently suspending a model’s development, deployment or internal use. A federal judge in the District of Columbia could order a suspension.
This approach would establish a defined threshold for intervention rather than giving government agencies unrestricted authority to halt AI development.
Anthropic’s Existing Safety Framework
An Anthropic spokesperson told Reuters that the proposal reflected measures outlined in the company’s Advanced AI Framework, which was published earlier in 2026.
The framework sets out proposals for governing advanced AI systems, including testing for serious risks, independent assessments, ongoing disclosure and government accountability. Anthropic has also advocated giving governments legal authority to prevent dangerous deployments.
The US Department of Commerce did not immediately respond to Reuters’ request for comment on the reported bill.
The proposal comes amid wider debate over how governments should regulate frontier AI models while allowing development to continue. The draft’s reported provisions would establish specific conditions under which authorities could seek to stop a model from being released or used.
