Anthropic is pitching federal lawmakers on draft legislation that would set strict conditions for when the US government can halt an advanced AI model, according to a confidential draft bill obtained by POLITICO.
Under the proposal, federal authorities could seek to stop a frontier model only when it presents a foreseeable, imminent threat - defined as qualifying risks involving more than $1 billion in damages or the deaths or serious injuries of over 50 people - and any suspension would require judicial approval rather than agency action alone.
Anthropic says the draft reflects its Advanced AI Framework, which it has promoted to federal and state lawmakers since June. "These are sensible policies we believe Congress should act on," an Anthropic spokesperson said of the effort.
The proposal would give frontier AI developers a clearer legal framework while narrowing the circumstances that could trigger government intervention. It also lands amid an active fight over who regulates AI in the US: the thresholds would place much of the enforcement burden on courts and on proving catastrophic risk in advance, a bar critics of light-touch rules say is too high and one safety advocates for pre-emptive oversight say defeats the purpose.
The draft is not introduced legislation, and there is no bill number, sponsor list or schedule. What exists is a lobbying artifact: a company-authored template for how it would like federal frontier-AI intervention powers to be constrained.