Trump Signs Executive Order on AI Model Reviews

President Donald Trump signed an executive order titled "Promoting Advanced Artificial Intelligence Innovation and Security" on June 2, 2026, according to the White House. The order establishes a voluntary framework under which developers of the most capable, or "frontier," AI models can give the federal government early access to those models for up to 30 days before releasing them more widely, so agencies can evaluate cybersecurity risks (White House text; Reuters; NPR). Reporting by The New York Times and Reuters frames the order as a shift away from the administration's previously lighter-touch posture, and The New York Times describes it as the administration's most significant AI policy action to date.

The order directs the Secretary of the Treasury, in consultation with the National Cyber Director and the directors of the NSA and CISA, to form an "AI cybersecurity clearinghouse" within 30 days. Legal analyses (Ropes & Gray, CFR) describe the clearinghouse as a voluntary mechanism to coordinate and deconflict scanning for software vulnerabilities, validate them, and prioritize remediation and patch distribution alongside industry and critical-infrastructure operators. The order also contemplates that participating developers will work with the government to select "trusted partners" granted early access to covered frontier models, though analysts note it provides no explicit criteria for choosing those partners.

The order explicitly states it does not authorize a "mandatory governmental licensing, preclearance, or permitting requirement" for the development, release, or distribution of AI models, language emphasized in both the White House text and subsequent legal commentary. Multiple outlets, including The New York Times and Reuters, reported that an earlier draft set the government-access window at 90 days; the reduction to 30 days was, according to legal analysts, the most significant change in the final order and reflected a compromise between national-security and anti-regulation factions within the administration.

As a general pattern, voluntary government-review frameworks tend to be shaped less by their formal text than by the follow-on agreements that define scope, confidentiality, and timelines. Several legal analyses published after the order, such as Ropes & Gray's, argue that a nominally voluntary program can still carry practical implications if participation becomes a de facto expectation for vendors selling to government or operating in critical sectors; other commentators emphasize that the order deliberately avoids a statutory licensing regime. Both readings turn on implementation details that are not yet settled.

• •Whether major developers sign voluntary testing agreements, and how "covered frontier models" and "trusted partners" are defined in practice.
• •How the Treasury-led clearinghouse operates: the data it requires, its retention and sharing policies, and how it interfaces with CISA, NSA, and industry.
• •The benchmarks agencies adopt for assessing models' cyber capabilities, and whether congressional or state action follows the order.