Bipartisan coalition pushes Trump administration on AI security framework release

A group of organizations from both sides of the aisle called on President Trump to publicly release his administration’s framework on government AI testing, arguing the closed-door policy raises concerns around transparency.

“The government has a fundamental responsibility to share the parameters of the process for reviewing frontier AI models prior to wider deployment,” the organizations wrote in the Wednesday letter . “Through this framework, the federal government will obtain initial access and visibility into the most transformative technologies we have ever seen.”

The letter, first reported by Semafor , was led by the Center for Democracy and Technology and Americans for Responsible Innovation, a nonprofit pushing for tighter guardrails on AI development.

The technology community was thrown off guard last month after the White House completed but decided not to release a long-awaited framework for how the government plans to test the security of private AI models ahead of their release.

The framework was a required part of Trump’s June executive order, which officially launched a new process for AI companies to voluntarily share their models with the government for up to 30 days before releasing them publicly.

While the order gave multiple agencies 60 days to develop a classified benchmarking process that would stay behind closed doors, it also called for a framework on how a company can be designated a “covered frontier model” and collaborate with the federal government to select “trusted partners” that also have early access to new models.

Among the signatories are the left-of-center consumer rights group Public Citizen, the Tech Oversight Project and conservative-leaning groups like the R Street Institute and Americans for Prosperity.

The organizations argued that while there may be legitimate reasons to withhold some portions for confidentiality or national security purposes, “the government has an unequivocal duty to disclose information about the framework’s standards, requirements, structure, and operational details.”

The letter warned this could decrease public trust in the government’s oversight of advanced AI, while disadvantaging smaller firms.

Representatives from Anthropic, Meta, OpenAI, Microsoft, Nvidia and some other smaller companies attended a briefing from the White House on the framework, but other parts of the AI industry were not looped in.

Several groups have pushed back against the White House’s decision in recent weeks. The nonprofit organization Protect Democracy sued the Trump administration earlier this month to enforce a Freedom of Information Act (FOIA) request asking for information about the framework.

The Foundation for American Innovation, a center-right think tank focused on technology and policy, also filed a FOIA request to the White House’s Office of the National Cyber Director for the release of the framework last month.

The framework comes at a time of intense scrutiny about the security of AI models and the role governments should play in their release.

OpenAI in July revealed its agents hacked into the systems of open-source tech firm Hugging Face, sending shock waves through the industry.

Anthropic and Meta later disclosed other incidents in which a misconfiguration by the cybersecurity company Irregular allowed models in isolated environments to access the internet and hack other systems.