What we know

Anthropic has proposed an AI safety bill that mandates reporting and regulatory steps only if an AI system poses an imminent risk of causing at least 50 deaths or $1 billion in damages. Under the proposal, evaluators must notify the Department of Commerce within 72 hours of identifying such risks. Following this notification, the Department is required to present the evidence to a federal judge for further action. The bill triggers regulatory action only when an AI system threatens 50 or more deaths or $1 billion in damages. Evaluators have a 72-hour window to report imminent risks to the Department of Commerce. The Department of Commerce must bring the evidence before a federal judge after receiving the report.

Why it matters

Anthropic’s proposal introduces a legislative framework aimed at AI safety with a notably high threshold for intervention. By requiring a substantial risk level—either mass casualties or major financial damage—before regulatory measures activate, the proposal could limit oversight to only the most severe AI risks.

What is still unknown

Details about how the bill would be enforced in practice remain unclear. ” There is no information on whether the bill covers all AI systems or only certain categories. No independent analysis of the proposal’s potential effectiveness or impact has been published.