1 of 1
Story summary
- Anthropic submitted a draft bill to Congress aimed at regulating frontier AI safety.
- The bill would let government block a model that could cause $1 billion in damages or kill over 50 people.
- The bill requires testing, disclosure, independent evaluation, evaluator registration with Commerce, and defines risks from bioweapons, control evasion, or cyberattack.
- Commerce may ask a D.C. federal judge to suspend a model, with no other agency authorized.