Anthropic drafts AI safety bill allowing model block if 50+ people killed or $1B damage risk

Anthropic has proposed a draft AI safety bill that would permit the blocking of a model release if an independent evaluator identifies a foreseeable, imminent risk of over 50 people being killed or more than $1 billion in property damage. The risk must stem from the creation of a biological weapon, the model evading developer controls, or a cyberattack. Under the proposal, the Commerce Department would need court approval to halt distribution, explicitly stripping federal agencies of unilateral authority.