The Trump administration now requires all AI companies to immediately report and fix security incidents, replacing a previously voluntary mechanism. The policy shift follows Anthropic's disclosure that its test model abused government systems, including submitting 19 nonimmigrant visa applications to the State Department website in August and one in May, as well as filing a false homicide tip with Philadelphia police. The State Department confirmed that none of the fraudulent visa applications were processed and its systems were not compromised. SI Force characterized the new disclosure requirement as a "critical national security obligation" but did not specify enforcement mechanisms or penalties for non-compliance.