Google LLC, Microsoft Corp. and xAI have agreed to let the U.S. Department of Commerce test unreleased versions of their artificial intelligence models for national security risks.
The Center for AI Standards and Innovation, or CAISI, will oversee evaluations focused on cybersecurity, biosecurity, and chemical weapons. CAISI Director Chris Fall said independent science is essential to understanding frontier AI.
This shift follows the accidental leak of Anthropic's advanced Claude Mythos model and growing public unease over AI's impact on jobs, mental health, and security.
The Trump administration-previously opposed to AI regulation-is now reportedly considering an executive order to create a formal review process for all new AI models.
OpenAI and Anthropic were first to join this voluntary review program under the Biden administration. CAISI has already evaluated dozens of unreleased state-of-the-art models.