IMPLICATOR
The deadline President Donald Trump set expired on August 1 with nothing published. The White House had not shown the public the new model-review framework he ordered.
The framework had been finished.
The White House acknowledged its completion on Monday, August 3, while keeping the document from public view. Its contents would guide how AI companies seek government review, even though the two measures that decide which models face scrutiny are classified.
What Changed
- The White House said Monday, August 3, that the voluntary AI model-review framework required by Executive Order 14409 was finished by its August 1 deadline, but it will not release the document or say who has seen it.
- The order classifies the cyber benchmarking process and the threshold that decides which models count as "covered frontier models," while Section 3(b), which creates the framework itself, carries no classification designation.
- Participating developers may give the Federal Government access to a covered frontier model for up to 30 days before releasing it to other trusted partners; Section 3(c) bars any mandatory licensing or preclearance requirement.
- A staff-level meeting with the Office of the National Cyber Director is set for Tuesday, August 4, with OpenAI, Anthropic, Google and Meta expected to attend.
AI-generated summary, reviewed by an editor. More on our AI guidelines.
The unpublished framework
“The voluntary framework outlined in the June 2nd executive order was complete by the deadline,” a White House official told Axios. Trump signed Executive Order 14409 on June 2, 2026, giving federal officials 60 days to finish the process.
Under Section 3(a), the order classifies the benchmarking process used to assess advanced cyber capabilities and the threshold that determines whether an AI model is a “covered frontier model.” That designation means a model powerful enough to enter the government’s early-review process.
That determination falls to the NSA director after consulting the National Cyber Director, the president’s science and technology adviser, the CISA director and other defense representatives as appropriate. Assessments may be shared with developers and researchers.
Section 3(b) orders the voluntary framework but does not designate that document as classified. The White House is withholding it anyway. “Just because things are unclassified that doesn't mean we are going to broadcast them to everyone,” the official explained.
Participating developers may provide the Federal Government with access to covered frontier models for up to 30 days before they plan to release such models to other trusted partners. The arrangement is supposed to include protections for confidential information, cybersecurity, insider risk and intellectual property. Section 3(c) says it cannot authorize “a mandatory governmental licensing, preclearance, or permitting requirement.”
The companies at the table
OpenAI, Anthropic and Google gave the administration feedback on a draft before the August 1 deadline. The White House official said that discussions about next steps were underway and that officials were engaging with “many more” industry partners than those three.
A staff-level meeting with the Office of the National Cyber Director is scheduled for Tuesday, August 4, Politico reported, citing five people familiar with the planning. Meta is also expected to attend, according to National CIO Review. CNBC reported, citing a source familiar with the plans, that Anthropic representatives are expected. The Information reported that OpenAI and Google are also likely to participate.
Get Implicator.ai in your inbox
Strategic AI news from San Francisco. No hype, no "AI will change everything" throat clearing. Just what moved, who won, and why it matters. Daily at 6am PST.
No spam. Unsubscribe anytime.
The White House has not said who has seen the completed framework. The administration has not said whether it will publish the document or when companies will start using it. The benchmark and qualifying threshold are classified, although the executive order says assessments and the threshold may be shared with AI developers and researchers as appropriate. OpenAI, Anthropic and Google have offered no on-record comment about the finished process.
OpenAI disclosed last month that an experimental agent escaped a restricted testing environment and compromised Hugging Face’s systems while trying to obtain answers for a cybersecurity evaluation. Hugging Face CEO Clément Delangue told CNBC on Monday that the breach showed the growing risks from increasingly autonomous AI systems.
The July warning
Brad Carson, president of Americans for Responsible Innovation and a former U.S. congressman and former acting under secretary of defense, joined Brendan Steinhauser, chief executive of the Alliance for Secure AI, in organizing a July 30 letter to Trump.
Know someone who'd find this useful? ✉️ Email it to a friend in one click, or they can subscribe free here.
The letter carried the names of 15 AI policy, safety and research leaders. It was copied to Acting Attorney General Todd Blanche and National Cyber Director Sean Cairncross. Signatory organizations included Public Citizen, the Future of Life Institute, FAR.AI and Palisade Research. The letter asked the administration to open an independent investigation, supported by outside auditors, into the OpenAI and Hugging Face breach. “We could not have asked for a clearer warning shot,” the signatories wrote.
They argued that the government “has a limited view into the advanced capabilities of frontier AI models and has not yet released a clear process for evaluating the safety concerns posed by advanced AI systems.” The requested inquiry would cover how the breach occurred and whether existing safeguards and reporting mechanisms were adequate. It would also identify steps to prevent another incident.
The coalition sought a rules-based process following Trump’s June 2 order. Its letter called for transparency into model capabilities, testing and incidents, along with meaningful evaluation before deployment.
The appeal went out two days before the Section 3 deadline.
The next meeting
Tuesday’s meeting will put company representatives in a room with staff from the Office of the National Cyber Director.
The White House has disclosed no publication date. The coalition ended its July 30 letter with its own timetable: “The administration should act before a warning shot becomes a preventable disaster.”
Frequently Asked Questions
What did the White House actually announce?
That the voluntary framework required by Section 3 of Executive Order 14409 was complete by its deadline. A White House official told Axios the framework "was complete by the deadline," but the administration has not published the document, named who has reviewed it, or said when companies will begin using it.
Why is the framework not public if it is not classified?
The executive order classifies the benchmarking process and the covered-frontier-model threshold under Section 3(a), but it does not designate the framework created under Section 3(b) as classified. Asked about the gap, a White House official said: "Just because things are unclassified that doesn't mean we are going to broadcast them to everyone."
What is a "covered frontier model"?
A model the government judges powerful enough to enter its early-review process. Under Section 3(a), the NSA director makes that determination after consulting the National Cyber Director, the president's science and technology adviser and the CISA director. The threshold itself is classified and shared with developers and researchers only as officials deem appropriate.
Does joining the framework mean the government can block a model's release?
No. Section 3(c) states that nothing in the section authorizes "a mandatory governmental licensing, preclearance, or permitting requirement" for developing or releasing new AI models. Participation is voluntary, and the access window runs up to 30 days before a developer releases a model to other trusted partners.
Who is objecting, and on what grounds?
On July 30, fifteen AI policy, safety and research leaders organized by Brad Carson of Americans for Responsible Innovation and Brendan Steinhauser of the Alliance for Secure AI wrote to President Trump seeking an independent investigation into OpenAI's Hugging Face breach. They argued the government "has a limited view into the advanced capabilities of frontier AI models."
AI-generated summary, reviewed by an editor. More on our AI guidelines.
Marcus Schuler
Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: editor@implicator.ai
