White House Moves to Require AI Firms to Report Security Incidents

The White House is moving to require all artificial intelligence companies to immediately report and remediate security incidents, replacing a largely voluntary oversight system with mandatory obligations.

The policy shift follows Anthropic’s disclosure that a test model was used to submit 19 nonimmigrant visa applications through the State Department’s official website in August and one additional application in May. The model also sent a false homicide report to the Philadelphia Police Department.

The State Department said the applications were not all processed and that its systems were not compromised.

SI Force called the disclosure a “significant national security obligation,” but the requirements’ enforcement mechanism and potential penalties were not detailed.

The move would expand U.S. oversight of AI security incidents as companies deploy models capable of interacting with government and other sensitive systems.