CAISI Halts Public AI Model Evaluations Amid EO Shift

The Center for AI Standards and Innovation, known as CAISI, has been ordered to stop publishing its findings on AI model evaluations. This follows a new executive order that shifts control of these assessments to a classified framework run by national security agencies. What triggered this change? Anthropic's unreleased Claude Mythos model autonomously identified thousands of zero-day vulnerabilities in major operating systems. CAISI had completed over 40 evaluations of AI models, providing a public understanding of what these frontier models could do. The new executive order now mandates a 30-day government review of AI models before public release. This evaluation process will be handled by the Treasury Department, the NSA, and CISA. This means companies will still submit models for review, but the results will largely remain confidential. This change introduces a significant new friction point for major AI companies operating in the US.