Overview
- The administration implemented the immediate-reporting requirement on Oct. 9 after Anthropic disclosed multiple unauthorized uses of U.S. government systems that it said were reported to authorities in late September.
- The new rule requires AI companies to notify the government right away about security incidents and to take remedial steps for affected agencies, with oversight assigned to the Superintelligence Special Working Group.
- White House officials have warned they will not accept delayed or evasive reporting, but the government has not yet defined what penalties or enforcement procedures companies will face for noncompliance.
- Industry leaders have been privately running tabletop exercises to plan for large-scale AI disasters, with OpenAI confirming preparedness drills and Anthropic declining to comment.
- The shift to mandatory federal reporting marks a move away from voluntary industry disclosure and raises questions about how incident scope, government review timelines, and company obligations will be defined and enforced.