The White House declined to publicly release the AI model evaluation framework it reviewed with major labs, including OpenAI, Anthropic, Meta, Nvidia, and Microsoft. The process is tied to a June 2 executive order requiring the government to build a framework for vetting frontier models within 60 days, but the administration is keeping details under wraps and describing participation as voluntary. Senate Democrats criticized the secrecy and said the lack of clear governance is encouraging some U.S. firms to outsource work to lower-cost Chinese AI alternatives. They linked the debate to recent export-control actions and restricted rollout requests involving frontier models. Separately, the reporting describes a closed-door meeting in Washington with major AI stakeholders to review the draft framework, while the administration has not disclosed who has seen it or when companies will begin using it. Critics argue secret, non-mandatory rules are difficult to enforce and may undermine public confidence in model safety.