AI TRENDS | Trump Administration Withholds Details of Voluntary AI Testing Framework

The Trump administration is refusing to release core documents on its voluntary AI testing framework, despite agreeing in September to disclose records in a Freedom of Information Act lawsuit. According to Sina Finance, citing The Information, which obtained the letter, the White House Office of the National Cyber Director sent a letter to the nonprofit Protect Democracy Project last Friday, invoking national security and commercial information exemptions to withhold the framework’s full draft and related records, including a list of organizations approved for early access to frontier AI models. The framework, established under a June 2026 executive order, allows companies such as Anthropic and OpenAI to submit new models for government review before public release. After finalizing it in early August, the White House briefed only a small group of companies, leaving others uncertain about key rules, including which organizations qualify as trusted partners with priority access before a model’s wider release. The letter’s document list describes the voluntary framework as a seven-page Word file; most other records are emails from the National Cyber Director’s office and the Commerce Department to AI labs and vendors regarding trusted partners’ pre-release access. Vivian Dong, a director at a technology security law advocacy group, said the public is being asked to trust a review framework whose details are known only to the government and AI companies. She said the group would continue its court fight until the full documents are made public. The White House did not respond to a request for comment.