Trump AI Executive Order 60-Day Framework Deadline Expires
Claire Weston
Trump's AI executive order, signed in early June, set August 1 as the deadline for federal agencies to finalize an AI regulatory framework. The White House says it will deliver on time. At its core is a classified benchmarking process that decides which models get regulated and who gets access — a mechanism set to reshape how AI companies ship products globally.
What does this framework actually regulate?
The order asks AI companies to voluntarily submit models for government evaluation before public release — voluntary, not mandatory.
The evaluation centers on a classified benchmarking process — an internal government scoring system, not publicly disclosed — focused on a model's cyber capabilities.
This means → the government is not trying to approve every AI model. It is drawing a single line: which models are powerful enough to be classified as "regulated frontier models."
The framework also encourages companies to work with the government to identify "trusted partners" — only organizations on that list would get access to the most capable models.
Who is driving the process, and who is lobbying?
Four officials are finalizing the plan: Treasury Secretary Bessent, Defense Secretary Hegseth, Chief of Staff Wiles, and Commerce Secretary Lutnick.
OpenAI CEO Sam Altman met Wiles and other officials in Washington this week. He told CNBC he had seen the draft and had "no particular edits to suggest."
Nvidia CEO Jensen Huang published an open letter urging policymakers to avoid "premature restrictions" on open-weight models — models whose parameters are publicly released for anyone to download. Microsoft, Meta, Palantir, and dozens of others co-signed; OpenAI later joined.
Anthropic did not sign the letter. It published a separate blog post laying out its own position. In plain terms = the top AI companies are not aligned on whether open-source models should be regulated.
The framework isn't final — but it's already having an effect?
Anthropic was forced to pull its Fable 5 and Mythos 5 models offline after a Commerce Department export-control directive.
OpenAI restricted the rollout of its GPT-5.6 series to "trusted partners" at the government's request.
Both companies were later cleared to release the models more broadly. This means → even before the formal framework is finalized, the government is already using the "spirit of the framework" to shape company behavior.
What is the biggest point of contention?
The fiercest policy divide ahead of the deadline: should Chinese open-weight models be restricted or banned?
Former AI czar David Sacks has been one of the loudest voices against any form of ban, arguing against a blanket approach.
The Trump administration has not taken a formal position. This reflects a deeper split: even within the executive order's own framework, the "security first" and "open competition" camps have not reached consensus.
Put simply = whether the framework lands on time on August 1 — and whether it clears up the regulatory gray zone around model deployment — is the key milestone to watch next.
Content is for reference only, not financial advice.