Trump Signs AI Executive Order Requesting Voluntary 30-Day Government Review of Frontier Models

--- headline: "Trump Signs AI Executive Order Requesting Voluntary 30-Day Government Review of Frontier Models" slug: trump-ai-executive-order-voluntary-review category: policy story_number: "14" date: 2026-06-02 ---

# Trump Signs AI Executive Order Requesting Voluntary 30-Day Government Review of Frontier Models

President Trump on Tuesday quietly signed an executive order asking AI companies to voluntarily hand over their most powerful models for government cybersecurity testing up to 30 days before public release -- a scaled-back measure that explicitly bars mandatory licensing but signals the administration is no longer willing to take a purely hands-off approach to frontier AI.

The order, titled "Promoting Advanced Artificial Intelligence Innovation and Security," was signed privately at the White House without the ceremony or livestream that typically accompanies major policy actions. There was no stage, no tech CEO photo op, no applause line. The subdued rollout reflects the political tightrope the administration walked to get here: satisfying national security hawks who wanted government oversight of the most capable AI systems while reassuring an industry that feared regulatory drag on the race against China.

From 90 Days to 30: The Lobbying Blitz

The final order is substantially narrower than the version Trump was expected to sign on May 21. That earlier draft proposed a 90-day government review window -- a timeline national security officials had pushed for. But hours before the planned signing ceremony, Trump pulled the plug.

"I think it gets in the way of, you know, we're leading China, we're leading everybody. And I don't want to do anything that's going to get in the way of that lead," Trump said at the time.

What happened between that scrapped ceremony and Tuesday's signing involved an intense lobbying campaign. According to multiple reports, SpaceX CEO Elon Musk and Meta CEO Mark Zuckerberg called Trump directly to oppose the original version. Former White House AI czar David Sacks -- who left his official role in late March but continues to co-chair the President's Council of Advisors on Science and Technology -- secured the shorter 30-day window, the voluntary framework, and explicit anti-mandatory-licensing language alongside National Economic Council deputy director Ryan Baasch.

Industry executives had lobbied for a review window closer to two weeks. The 30-day compromise represents a middle ground between that preference and the 90-day timeline the national security establishment sought.

Notably, the industry was not unified. OpenAI's chief lobbyist Chris Lehane had been broadly supportive of the original government-collaboration model, placing the company in direct opposition to Musk and Zuckerberg on the question.

What the Order Actually Does

The executive order tasks the Treasury Department, the National Security Agency, and the Cybersecurity and Infrastructure Security Agency (CISA) with developing benchmarks to evaluate the cyber capabilities of AI models. Within 60 days, those agencies -- along with the National Institute of Standards and Technology and White House officials -- must create a classified benchmarking process and decide when a model qualifies as a "covered frontier model."

Once that threshold is established, developers can voluntarily provide the government access to qualifying models up to 30 days before releasing them to "trusted partners." Companies can also work with agencies to select which outside partners get early access.

The order also directs the creation of an "AI cybersecurity clearinghouse" within 30 days -- a voluntary collaboration between industry and critical infrastructure operators to coordinate vulnerability scanning, validate discovered flaws, and prioritize patch distribution.

Additional provisions direct the Committee on National Security Systems and the Defense Department to prioritize cyber defense of their information systems, and task CISA with releasing Binding Operational Directives to expand federal use of AI-enabled defensive tools.

The anti-regulatory guardrail is stated plainly in the text: "Nothing in this section shall be construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement for the development, publication, release, or distribution of new AI models, including frontier models."

The Voluntary Problem

The central tension is obvious: a safety regime that depends on which CEO decides to cooperate is not, by most definitions, a safety regime. The government can ask. It cannot compel.

Sen. Josh Hawley, R-Mo., expressed support but said he would go further. "I think we ought to enact my legislation that I have with Senator Blumenthal that would make that sort of reporting and monitoring mandatory," Hawley said, referencing a bipartisan bill introduced last September that would require the Energy Department to conduct mandatory reviews of advanced AI systems.

Brendan Steinhauser, CEO of the Alliance for Secure AI, was more direct: "Congress must now codify the White House's EO with legislative action. Lawmakers need to create a legal framework that makes federal government review of advanced AI models mandatory."

Brad Carson, president of Americans for Responsible Innovation and a former House Democrat from Oklahoma, echoed the call for codification, calling recent AI-fueled vulnerabilities a "massive wake-up call" and describing the order's provisions on vulnerability detection and benchmarking as "positive steps forward."

A former Trump White House official, speaking to The Hill last month, raised perhaps the most pointed question about the voluntary structure: "What is to say it stays that way?"

Context and What Comes Next

The order emerged against a specific backdrop: Anthropic's announcement of its advanced cybersecurity model Mythos earlier this spring, whose ability to identify decades-old security flaws raised alarms that adversaries could exploit similar capabilities to target vulnerabilities in critical systems.

For an administration that revoked Biden's 2023 AI executive order on its first day in office and has championed deregulation as a competitive strategy, any government review mechanism -- even a voluntary one -- represents a meaningful philosophical shift. The question now is whether voluntarism proves sufficient, or whether the next AI incident forces the mandatory framework that Congress is already drafting in the wings.

The 60-day clock for benchmarking starts now. By early August, the government will need to define what counts as a frontier model. Only then will anyone know which companies are being asked to share -- and whether any of them say yes.