Anthropic pitches US AI bill with $1 billion, 50-injury trigger

Anthropic is circulating a draft federal bill to US lawmakers that would create a legal process for stopping the release or use of the most powerful AI models in narrowly defined high-risk cases. Under the proposal, the government could seek a court order if a model posed a foreseeable, imminent risk of more than $1 billion in damage or of killing or seriously injuring more than 50 people. The draft ties those thresholds to threats involving biological weapons, a model evading its developer’s controls or a cyberattack. It would also require frontier AI developers to test models, disclose findings, submit to independent evaluations and maintain security programs, as Congress still has not passed a federal AI law.