White House Finalizes AI Model Evaluation Framework Behind Closed Doors
The White House has finalized a voluntary framework for evaluating new AI models, granting the U.S. government and trusted partners 30-day access to AI systems before their public release. The details of the framework, its timeline, and the list of approved evaluators have not been published.
Experts had anticipated a transparent structure for developer-government interaction, including requirements for cybersecurity, intellectual property protection, and confidentiality during audits. According to sources, the White House plans to discuss the new standards with AI company representatives at an upcoming closed-door meeting.
Related: Singapore Launches Model AI Governance Framework for Agentic AI