Anthropic quietly pitches draft frontier AI safety bill to US lawmakers

Anthropic circulated a draft bill that would allow the US government to block release or use of a frontier model only under narrow conditions of imminent risk exceeding $1 billion in damages or 50 deaths/serious injuries from specified threats. The proposal requires testing, transparency, and independent evaluation while limiting broader federal intervention powers. This represents a major AI developer actively shaping federal legislation on frontier model oversight.