AI3 min
Anthropic backs AI safety bill that halts models only if 50 deaths or $1B damage loom
Tom's Hardware, citing Politico, reports that Anthropic is lobbying for a federal AI safety bill. The government could block a model's release only if it risks $1 billion in damages or killing or injuring more than 50 people, from bioweapons, AI-enabled cyberattacks or a lab losing control of a model. An independent evaluator registered with the Commerce Department would check labs and report risks within 72 hours. A federal judge in Washington, DC would then decide whether to halt the release. No other agency could order a lab to stop. An Anthropic spokesperson said the bill reflects its Advanced AI Framework and called these 'sensible policies'.
Hype check: This is a lab's own proposal, not a law. Critics quoted in reader comments call it regulatory capture, because it would leave only one agency able to stop a model.