The Pulse
Anthropic Pitches Lawmakers a Draft Bill to Restrict AI Shutdowns
Anthropic representatives pitched lawmakers a draft bill that would limit when the federal government could block the release or use of frontier AI models. The reported proposal includes independent evaluations and would make a court order
AI.info Team · Written with AI ·
Anthropic representatives have quietly pitched Capitol Hill lawmakers on a draft bill to regulate frontier AI safety and security, POLITICO reported on October 16, citing three people with knowledge of the effort. Reuters reported that the proposal would let the government 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.
A high bar for government intervention
The draft ties those thresholds to three kinds of danger: creating or releasing a biological weapon, a model evading its developer’s controls, or a cyberattack. The reported conditions would limit intervention to specified risks and consequences, rather than give the government a general power to halt a model over concerns about its capabilities.
Under the proposal, the Commerce Department could ask a federal judge to suspend a model after determining that it poses the defined risk. A judge in the U.S. District Court for the District of Columbia could order the suspension. The draft would bar other federal departments, agencies or officials from suspending a model’s development, deployment or internal use, according to Reuters’ account of POLITICO’s reporting.
Evaluators would report urgent findings to Commerce
An Anthropic spokesperson told Reuters that the bill requires developers to test frontier AI models, be transparent about findings and get them independently evaluated. Evaluators would register with the Commerce Department and periodically assess developers and their frontier models. Companies could pay for evaluations, but payment could not depend on the evaluator’s conclusion, Reuters reported.
If an evaluator found what the draft calls an “imminent enumerated risk,” it would have to notify Commerce within 72 hours of completing its assessment. The Commerce Department could then request that a judge halt a model after determining that it poses the defined risk, according to Reuters’ account of POLITICO’s reporting.
Anthropic links the draft to its June framework
The spokesperson told Reuters that the bill represents proposals outlined in Anthropic’s Advanced AI Framework and would give the government the “ability to block or deter dangerous deployments of the most powerful models.”
Anthropic’s June 2026 Advanced AI Framework discusses model testing, external evaluation and government authority to restrict dangerous deployments. The framework’s authors write:
Ultimately, there should also be a way to block or deter deployment of models that pose significant catastrophic risks.
The June framework presents company recommendations for government action and discusses safeguards on regulatory authority, including court enforcement as one option. It proposes that an agency could pursue remedies by filing a lawsuit, while allowing provisional remedies in cases of imminent catastrophic risk. The draft bill described in POLITICO’s reporting would make a court order central to the route for suspending a model.
A private pitch, not enacted legislation
The measure remains a draft circulated among lawmakers, not a law. The bill would also preempt some state AI safety laws, POLITICO reported. The proposal pairs mandatory evaluations and reporting with a narrow federal trigger for blocking a model, while leaving the final suspension decision to a judge.