Bipartisan coalition urges White House to disclose its AI safety framework

Over twenty organizations representing both liberal and conservative interests have signed a letter demanding that the White House publish its voluntary AI security framework. The appeal, spearheaded by Americans for Responsible Innovation and the Center for Democracy and Technology, argues that the government must disclose the parameters guiding the review of frontier AI models prior to broader deployment. The framework was initially shown last month to a select group of technology companies, but smaller and open-source firms were largely excluded.

Recent events, including OpenAI’s assault on the HuggingFace repository and the rollout of its Astra model, have intensified calls for openness. Protect Democracy has also filed a lawsuit to compel the administration to honor a Freedom of Information Act request for the framework, questioning the legal basis of the original Trump-era order. The signatories warn that a lack of transparency could undermine public trust and entrench dominant AI labs while leaving open-source developers in the dark, urging congressional involvement as well.