Anthropic Circulates Draft AI Safety Bill to Lawmakers

Anthropic is circulating a draft federal bill on frontier artificial intelligence safety among U.S. lawmakers, proposing independent assessments of advanced models and sharply limiting when the government could halt their release or use. The proposal has not been formally introduced, and its provisions remain those of a draft. According to accounts of the draft, federal intervention would be available only for a foreseeable, imminent risk involving more than $1 billion in damages or the deaths or serious injuries of more than 50 people. The risks would have to fall within specified categories: biological weapons, a model escaping its developers’ controls, or cyberattacks. The proposal would require registered independent evaluators to assess frontier AI developers and their models periodically. Companies could pay for the evaluations, but compensation could not depend on the findings. Evaluators who identify an imminent risk covered by the draft would have to notify the Commerce Department within 72 hours after completing an assessment. The Commerce Department could then ask a federal judge to order a model halted. Other federal departments, agencies and officials would be...