Anthropic’s proposed AI safety bill would tie federal power to block an advanced AI model to thresholds for severe harm. POLITICO’s October 21 account draws a distinction with California Senate Bill 53: California uses similar figures for companies’ risk assessments and disclosures to state officials, while Anthropic’s draft would make them a condition for federal model-blocking authority.
POLITICO reported on October 16 that Anthropic representatives had privately pitched the draft to lawmakers. Its October 21 follow-up continued to describe a proposal, not an enacted law. The broader debate over external AI oversight also appears in NeoTeo’s earlier coverage of Sarah Heck, Dario Amodei and Jensen Huang’s views on AI safety.
What the reported threshold covers
POLITICO’s October 16 report described a “foreseeable, imminent” risk of more than 50 people being killed or seriously injured, or damage exceeding $1 billion. Its October 21 follow-up used different wording: 50 or more people, or $1 billion in damage. The October 16 account listed biological-weapon creation or release, a model evading its developer’s controls, and a cyberattack as qualifying risk scenarios.
How the proposed suspension process would work
A Business Insider republication of POLITICO’s reporting described independent evaluators registered with the U.S. Department of Commerce, who would periodically assess AI labs and their frontier models. Labs could pay for an evaluation only if payment did not depend on its outcome. An evaluator identifying an imminent enumerated risk would have 72 hours after completing the assessment to report it to Commerce.
Commerce could then ask a federal judge in the District of Columbia to halt a model. Other federal departments, agencies and officers would be barred from suspending its development, deployment or internal use under the reported draft.
Other proposed requirements
The draft would require labs to maintain safety frameworks, publish transparency reports for new frontier models, protect whistleblowers and report critical safety incidents to Commerce. It would also preempt some state requirements covering public risk disclosures, third-party evaluations and critical-incident reporting.
