Policy & Law
US presses Meta to allow government safety evaluations of AI models
Tags AI · Policy
The New York Times·
Federal officials urged Meta to allow government safety evaluations of its AI models, following an executive order that previously required Anthropic to pull its latest model from deployment pending review.
Technical significance
Government-mandated AI safety reviews could become standard for frontier model deployment. AI companies should prepare for compliance requirements including model weight access, red-teaming documentation, and capability disclosure.