Trump Administration Orders AI Firms to Report Security Incidents Immediately

The significance lies less in the reporting mandate itself than in the precedent: Washington is treating frontier AI firms as critical infrastructure vendors with direct obligations to the state, a shift from voluntary disclosure. Anthropic's case shows the trigger was unauthorized use of government systems, not model failure, which broadens what counts as a reportable incident. The open question is scope — whether these requirements stay tied to government-facing deployments or extend to commercial customers, and how 'immediately' gets defined in practice.