White House withholds AI safety framework, drawing criticism

The White House declined to publicly release its voluntary AI safety framework, drawing criticism from watchdogs and lawmakers who say secret oversight lacks accountability. The framework was meant to clarify which AI systems the government reviews before public release; leaked parts say it focuses on closed models. White House summoned OpenAI, Anthropic, Meta and Google to discuss it. The U.S. lacks AI regulation statutes, and critics say secrecy favors large incumbents. Lawmakers called for civilian oversight; concerns include Chinese open-weight models and a broad trust gap.