Civil society groups press US to disclose frontier AI review framework
A coalition of civil society and advocacy organisations is urging the US administration to make public the framework it is developing to review frontier AI systems, arguing that confidential oversight could weaken accountability and disadvantage smaller and open-source developers.
A coalition of organisations has called on the US administration to publicly release the voluntary framework being developed to review frontier artificial intelligence systems under the White House’s 2 June executive order.
In a 9 September letter to the president, the groups said the framework should be developed and implemented transparently. They argued that although some information could legitimately be withheld for confidentiality or national security reasons, the government should disclose the framework’s standards, requirements, structure and operational details.
The organisations raised concerns about a process under which the federal government would receive priority access to frontier AI models before wider deployment. They argued that such access gives the government significant visibility into increasingly powerful technologies and could affect whether and how they become available to the public.
The letter also warned that limiting access to the framework could have implications for competition in the AI sector. According to the groups, a process shared primarily with the largest AI companies could disadvantage smaller and open-source labs, influence market structures and contribute to regulatory capture.
The coalition further called for Congress to have access to the framework as it considers legislation on AI. It said transparency would allow lawmakers to better assess the administration’s approach as they work on potential safeguards for frontier AI.
The groups framed public oversight as a central requirement for the governance of advanced AI. They argued that a lack of accountability over the process could undermine public confidence in government oversight.
The letter concludes by calling for greater public discussion of the framework, while linking transparent oversight to the administration’s stated objectives of maintaining US leadership over China in AI and advancing innovative and safe frontier AI systems.
The letter was signed by 31 organisations, including the Center for American Progress, Center for Democracy & Technology, Common Cause, Consumer Federation of America, Fight for the Future, Public Citizen, Public Knowledge, R Street Institute and the Tech Oversight Project.
