Sens. Elizabeth Warren and Richard Blumenthal are asking the Trump administration to disclose who helped shape its unpublished AI review framework, how the government would test models and what could happen if a model fails an evaluation. Their request puts the framework’s development under scrutiny, but the available reporting does not establish whether officials have answered their letter or whether the framework has been used.
In a letter dated Monday, the Democratic senators asked six administration officials for records and details. The recipients include Treasury Secretary Scott Bessent, National Cyber Director Sean Cairncross, Commerce Secretary Howard Lutnick and White House Chief of Staff Susan Wiles. The letter requests a response by Oct. 19.
The senators’ five numbered requests cover records of industry-government meetings and communications about Executive Order 14409 and its drafts; details of an industry questionnaire; whether people outside the industry were consulted; and an account of the framework’s testing process, including what follows a failed evaluation. Semafor reported on the letter.
Warren and Blumenthal said in the letter, as reported by Semafor, “We are concerned that the Trump Administration is accommodating the interests of Big Tech CEOs rather than addressing the safety and security risks posed by increasingly sophisticated AI systems.” That is the senators’ concern, not an independently established finding that industry interests shaped the framework.
What the framework is supposed to do
President Donald Trump’s June 2 executive order directs agencies to create a classified process for benchmarking advanced AI cyber capabilities and to design a separate, voluntary framework for government review of covered models. Under the order, developers could give the government access to a covered model for up to 30 days before releasing it to trusted partners.
The distinction matters: the order classifies the benchmarking process, but the cited reporting does not establish that the entire framework is classified. Axios reported that the order does not explicitly require the framework to be published and does not designate the framework itself as classified.
Free newsletter
Get the morning briefing
Start each day with the AI stories that matter and why — a short, free email from our newsroom.
Axios reported on Aug. 3 that a White House official said the administration completed the framework by its Aug. 1 deadline and that discussions with industry about next steps were continuing. The administration did not disclose the framework’s contents or when companies would begin using it in that account. Semafor reported on Aug. 5 that the administration had told a handful of firms it would not publish the framework; the White House did not respond to that outlet’s request for comment on why.
Those reports establish what the administration said about completion and publication, not how the review would work in practice. The available reporting does not identify which companies have opted in, show that any model has been tested under the framework, or describe its coverage criteria and evaluation methods.
What the senators’ questions could clarify
The requested records could show which companies or other groups had access to officials during development, whether outside experts were consulted and what standards the government plans to apply. The questions about failed evaluations could also clarify whether the voluntary process includes any consequence or follow-up. The letter asks for this information; the reporting does not establish what the answers will be.
That information bears on public oversight because the framework would guide government review of covered models while its terms remain undisclosed. A voluntary structure is not the same as a public rule with stated requirements. Without the framework’s details, the public cannot assess which models fall within its scope or how consistently the government would evaluate them.
Semafor’s Oct. 6 report does not say whether officials have responded to the new letter. The senators’ requested deadline is Oct. 19; the framework’s contents, any model testing and the administration’s response remain unknown in the cited reporting.