The bill would reportedly let the government block a model’s release or use only if it poses a foreseeable, imminent risk of more than US$1 billion in damages, or of killing or seriously injuring more than 50 people.
Anthropic quietly pitches its own AI safety bill to US lawmakers
