Anthropic pitches AI safety bill that may block a model if 50+ people could be killed

Anthropic has proposed a draft AI safety bill to the U.S. Congress that would block AI model releases posing imminent catastrophic risks, such as potential deaths over 50 or property damage exceeding $1 billion. Independent evaluators registered with the Commerce Department would assess risks, and the department would seek judicial approval to halt risky models within 72 hours. The bill aims to regulate AI safety with defined thresholds and an evaluation process separate from federal agency sole