Warren, Blumenthal seek records on White House AI framework
Senators Elizabeth Warren and Richard Blumenthal gave six administration officials until October 19 to turn over records and answer questions about industry involvement in the White House’s AI framework and its pre-deployment testing.
U.S. Senators Elizabeth Warren and Richard Blumenthal sent an October 5 letter to six administration officials, seeking records and answers about the White House’s AI framework by October 19.
Their five-part request covers who helped shape Executive Order 14409 and how the administration’s pre-deployment testing process works. A response could disclose private-sector contacts, changes made while the order was drafted and details of a classified model-evaluation process.
The senators want meeting records and communications with private-sector representatives, drafts and redlines showing material changes to the order, and the reported 31-item questionnaire from the Office of the National Cyber Director. They also asked for a list of questionnaire recipients and their responses.
Warren and Blumenthal alleged that industry pressure weakened the framework and that the administration favored technology executives over public safety. They also described the process as voluntary, industry-designed, secretive and lacking enforceable safeguards. The available records do not independently establish those allegations. The communications, redlines and meeting records that could substantiate them are among the materials the senators are seeking.
The June 2 executive order directs designated agencies to maintain a classified benchmarking process that assesses advanced cyber capabilities and determines when an AI system is a covered frontier model. It also directs agencies to design a voluntary framework with developers that can give the government access to covered models for up to 30 days before a planned release to other trusted partners. The order says that section does not authorize mandatory government licensing, preclearance or permitting for new AI models.
That policy framework is separate from the White House’s America.gov AI service for federal information, a public-facing administration initiative.
The letter also asks whether the administration consulted independent evaluators, civil society groups, consumer advocates, critical-infrastructure operators and academic experts on the order or the classified evaluation process. If so, the senators want the related communications and feedback.
On testing, Warren and Blumenthal requested the criteria used to decide which models are covered; the capability and risk areas tested; the benchmarks and evaluations used; and any action taken when a model fails an evaluation or presents an unacceptable risk.
The letter went to Treasury Secretary Scott Bessent, National Cyber Director Sean Cairncross, Commerce Secretary Howard Lutnick, Office of Science and Technology Policy Director Michael Kratsios, White House Chief of Staff Susan Wiles and Office of Management and Budget Director Russell Vought. The senators asked all six officials to provide the requested material no later than October 19, 2026.
More news

Trump, six AI companies sign voluntary AI safety accord

White House launches America.gov AI service for federal information

Florida seeks court limits on OpenAI model development and ChatGPT behavior
