What did Trump's AI executive order require by August 1, 2026?

Executive Order 14409 gave agencies 60 days from its June 2 signing to build a classified AI cyber-capability benchmark, a voluntary pre-release review framework for frontier models, and a federal cybersecurity hiring plan.

How is this different from the White House already deciding who gets access to frontier AI models?

That earlier, informal gatekeeping — reported by CNBC in July — happened outside any of the structures the executive order was meant to create. The formal 30-day pre-release review process it promised still doesn’t exist as of the deadline.

What is a "covered frontier model" under the order?

The order doesn’t fix a single definition — it directs agencies to build the classified benchmarking process that would decide which AI models are advanced enough to count, based on cyber capability. That threshold was one of the pieces still missing after August 1.

White House Misses Deadline for AI Frontier-Review Framework

Q: Did the government publish any of the required frameworks by the deadline?

No. As of August 1, no Federal Register notice, NIST or CISA publication, or OSTP statement had addressed any of the three requirements.

A 60-day deadline set by President Donald Trump’s June AI executive order passed on August 1 without the federal government publishing any of the frameworks the order required, according to reviews of the Federal Register and agency websites by Yahoo Finance and Tech Times.

What the order required

Signed June 2 and titled “Promoting Advanced Artificial Intelligence Innovation and Security,” the order told the Treasury Department, Homeland Security and the Pentagon to jointly build, within 60 days, a classified benchmarking process for judging how dangerous an AI model’s cyber capabilities are. In parallel, the same agencies — working with the National Institute of Standards and Technology — were to design a voluntary framework letting frontier AI developers consult with the government and hand over early access to a model for up to 30 days before its public release. A third deadline required the Office of Personnel Management to expand federal cybersecurity hiring pathways.

Silence past the cutoff

As of August 1, no Federal Register notice, NIST or CISA publication, or OSTP statement had appeared addressing any of the three requirements. The regulation leaves AI developers without clarity on what counts as a “covered frontier model” or how the promised 30-day pre-release review would actually work.

An informal process already in motion

The formal framework’s absence is notable because, informally, the White House had already begun deciding which firms get early access to frontier models from labs including Anthropic and OpenAI, CNBC reported in July — before any of the order’s official mechanisms existed. That leaves government gatekeeping running ahead of the AI safety structure meant to govern it.

The order followed a wave of AI-linked security incidents this year that pushed cybersecurity higher up the policy agenda in Washington.

What the order required

Silence past the cutoff

An informal process already in motion

Frequently asked questions