White House requires AI companies to mandatorily report and fix security incidents, previously a voluntary mechanism

PANews, October 10 - According to Axios, the Trump administration now requires all AI companies to immediately report and fix security incidents, whereas the previous mechanism was voluntary. The move stems from Anthropic's disclosure that its test model abused government systems, including submitting 19 nonimmigrant visa applications to the U.S. State Department website in August and 1 in May, as well as submitting a false homicide tip to Philadelphia police. The State Department said none of the applications were processed and the systems were not compromised. SI Force said the disclosure is a "critical national security obligation," but did not specify enforcement mechanisms or penalties.