Anthropic proposes AI safety bill with strict death threshold and judicial approval

Anthropic, a leading artificial intelligence company, has proposed a draft bill that sets strict conditions for blocking the release of AI models, according to reports from TechCrunch and others.

The proposed legislation would permit federal agencies to halt an AI model only if it poses a "foreseeable, imminent risk" of causing 50 or more deaths or over $1 billion in damages, with the additional requirement of federal judge approval.

The bill also suggests that independent evaluators, registered with the U.S. Commerce Department, would assess potential risks of AI models. These evaluators could be compensated by AI labs, provided their payment is not contingent on evaluation results.

Some experts have raised concerns that shifting decision-making power from expert regulators to judges could delay urgent actions needed to address AI risks, as highlighted by TechBuzz.

Anthropic has been advocating this framework since June and aims to influence AI regulation before Congress enacts its own rules. The company seeks to balance safety oversight with maintaining industry flexibility.

The bill remains a proposal and has not been enacted into law.