backOn June 2, 2026, President Trump signed an Executive Order1No Executive Order number has been assigned as of the time of the publishing of this alert. (“EO”) establishing an expanded federal framework for engagement with advanced artificial intelligence (“AI”) models prior to their public release. Citing “national security considerations that require coordinated action across executive departments and agencies,” the EO represents the second Trump administration’s most comprehensive action on AI governance to date. The EO follows heightened attention to frontier-model cyber capabilities. In a social media post, the White House stated that it was not conducting AI model oversight “as that level of government overreach would have chilling effects on free speech and innovation.”
Key Provisions
Most significantly, the EO establishes a voluntary framework under which private companies may engage with the government to provide access to cutting-edge AI models for capability testing up to 30 days before public release.
The EO also instructs the Secretary of Homeland Security to issue “Binding Operational Directives” within 30 days, covering three priority areas: (1) expediting and prioritizing the cyber defense of the federal government; (2) establishing or expanding federal programs and cybersecurity services that enhance AI-enabled defensive tools; and (3) facilitating federal agency access to frontier AI models. Those directives will provide additional detail on how the administration intends to implement the framework across executive departments and agencies.
The EO further directs the government to “develop and maintain a classified benchmarking process to assess the advanced cyber capabilities of AI models.” This provision indicates that at least part of the administration’s evaluation framework will operate outside public view.
White House Deliberations
Reporting indicates that the EO’s path to issuance involved internal debate. According to Politico, a May draft that reportedly contemplated a 90-day voluntary review period for advanced AI models was withdrawn after former White House “AI czar” David Sacks urged the president to reconsider.
Public reporting on the drafting process suggests that the administration weighed competing priorities, including innovation, national security, and competitiveness. The final EO - centered on voluntary pre-deployment engagement rather than mandatory oversight - appears to reflect an effort to balance those considerations.
Congressional Reaction
The EO drew prompt bipartisan reaction on Capitol Hill, reflecting growing congressional interest in AI oversight. Senator Mark Warner (D-VA) agreed with the administration’s commitment to pre-deployment testing and reaffirmed his support for designating the National Security Agency to “lead efforts to evaluate national security implications of frontier models in voluntary partnerships with frontier vendors.” Senator Josh Hawley (R-MO) offered broader support for the EO’s objectives but made clear that he views the voluntary framework as an insufficient response, stating: “I would go farther. 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’s reference was to his Artificial Intelligence Risk Evaluation Act, co-authored with Senator Richard Blumenthal (D-CT), which would establish a program within the Department of Energy to evaluate advanced AI systems and mandate developer participation. Senator Bernie Sanders (I-VT), who has previously advocated for a moratorium on datacenter construction, called on Congress to act because the “executive order is voluntary and does almost nothing to protect Americans.”
Key Takeaways
The EO indicates a significant step towards a federal framework for AI regulation and has several implications for companies operating in or near the frontier AI space. First, developers should monitor the Binding Operational Directives that the Secretary of Homeland Security is directed to issue within 30 days, as those directives are expected to further define the new framework. Second, companies developing advanced AI models may wish to consider whether and how to engage with the voluntary pre-deployment review process as implementation details emerge. Third, the bipartisan reaction on Capitol Hill suggests that the current framework may be an initial step rather than the endpoint of federal AI oversight. We will continue to monitor and update on developments.
