US AI Executive Order Postponed Amid Industry Concerns
Published on: May 22, 2026
On May 21, 2026, President Trump postponed the signing of a highly anticipated executive order aimed at establishing a voluntary framework for AI firms to share new models with the government before public release. Trump cited concerns about overregulation and the potential impact on America’s competitive edge against China as key reasons for the delay. Industry figures including Elon Musk, Mark Zuckerberg, and venture capitalist David Sacks reportedly influenced the decision, arguing that stringent oversight could stifle innovation. Multiple sources confirm that the proposed order was intended to enhance cybersecurity and national resilience in the face of increasingly powerful models—but was ultimately set aside due to internal and external pressure.
The executive order was expected to create a voluntary mechanism for pre-release model sharing between AI companies and federal authorities, providing the government with insight into new, potentially high-risk systems before public dissemination. The move followed growing national security concerns surrounding AI capabilities, particularly after Anthropic’s unreleased Mythos model raised alarms among security officials. However, key tech leaders reportedly persuaded the White House that such measures could undermine the United States’ competitive pace in the global AI race, particularly with China advancing rapidly.
Behind the scenes, reports suggest that tech CEOs—including those leading major AI firms—engaged in discussions with the president and his advisors in the days leading up to the planned signing. Industry support for voluntary testing and model guardrails remained, but the involvement of the Treasury Department in overseeing model security was questioned by some stakeholders, adding complexity to the agreement's terms. As one tech source described it, dissent over the draft’s specifics played a critical role in triggering the postponement.
Experts warn that this delay may further entangle AI policy in partisan disagreements and interagency complexities. With the absence of federal action, some lawmakers have begun exploring legislative alternatives. Representatives Jay Obernolte (R–California) and Lori Trahan (D–Massachusetts) are reportedly developing a bipartisan bill intended to preempt state-level AI regulations for a limited period, hoping to introduce clarity and uniformity nationwide before the Commerce Department’s AI safety center receives funding.
What to watch next: The White House’s National Cyber Director has hinted at additional AI-related security initiatives that could emerge in lieu of the executive order. Meanwhile, industry and lawmakers alike remain divided over the balance between innovation and oversight. The delay underscores the complex interplay between national security, public policy, and industry dynamics in shaping the United States’ AI governance framework.