Anthropic proposal would send AI safety risks to a federal judge

Summary

Anthropic's proposed AI safety bill would require evaluators to report imminent risks to the Department of Commerce within 72 hours; Tom's Hardware reports that Commerce would then present evidence to a federal judge and await an order. The proposal's threshold is 50 deaths or $1 billion in damages. This follows earlier reports that Anthropic had pitched lawmakers on a draft AI safety bill.

Timeline

1. Tom's Hardware

   18 Oct 2026, 11:25 UTCSource ↗

   Tom's Hardware reports that evaluators would have to alert the Department of Commerce to any imminent risk within 72 hours, after which the department would present evidence to a federal judge and await an order.