The White House is moving forward with a new framework designed to evaluate the security of advanced AI models, but officials have confirmed that the document will not be made public. According to reports, the administration intends to limit access to the framework to the specific technology companies involved in the development process.

This decision has raised questions among policy experts and international partners who argue that the lack of transparency could hinder broader understanding of how the U.S. intends to manage AI risks. Because the framework is voluntary, the administration’s approach relies on cooperation with industry leaders rather than formal regulation. By keeping the evaluation criteria internal, the White House effectively prevents independent researchers and outside policymakers from scrutinizing the standards being applied to these powerful systems.

While the White House held staff-level meetings with industry representatives this week to discuss the implementation of these policies, the decision to withhold the framework suggests a preference for private collaboration over public oversight. Critics note that because these AI models have global security implications, the secrecy surrounding the evaluation process may complicate efforts to establish international norms or consistent safety standards. The administration has not yet provided a detailed justification for why the framework must remain confidential, though the move aligns with a broader strategy of working directly with major tech firms to address safety concerns without imposing immediate federal mandates.