WASHINGTON — President Donald Trump abruptly pulled back from signing a planned executive order on artificial intelligence after expressing concern that the proposal could weaken the United States’ advantage in the fast-moving global race for AI, according to people familiar with the matter and public remarks from the president.
The decision, first reported by Axios and discussed in a recent DW News segment, came just hours before a White House event that had been expected to put the administration’s latest AI policy on display. Instead, Trump said he had “postponed” the measure because he did not like “certain aspects” of it and did not want to do anything that could get in the way of America’s lead over China and other competitors.
The move underscores the tension inside the Trump administration between two priorities that often collide: reducing the perceived risks of advanced AI systems and accelerating U.S. innovation so American companies can stay ahead in a competition that has become a top national-security concern.
A sudden reversal at the White House
According to reporting and broadcast analysis, the draft order would have required major AI companies to provide the federal government with early access to new models before public release. That would have given agencies an opportunity to test systems for dangerous capabilities, security vulnerabilities and potential cyber risks before they reached consumers or enterprise users.
Administration officials had framed the idea as a safeguard for a technology that is becoming increasingly capable and, in some cases, difficult to fully evaluate after deployment. The proposal was also expected to strengthen cybersecurity by giving federal experts more visibility into how frontier models behave.
But Trump appeared uneasy about anything that might slow the private sector or add another layer of federal review. Speaking to reporters, he argued that the U.S. is already leading in AI and said he did not want to take action that could hinder that momentum.
“I didn’t like certain aspects of it,” Trump said, adding that he had postponed the order because he did not want to “do anything that’s going to get in the way” of America’s AI lead.
Why the order raised alarm
The draft executive order was reportedly motivated in part by growing concerns that the newest generation of AI systems is becoming powerful enough to pose real-world risks. Those concerns include the possibility that advanced models could assist with sophisticated cyberattacks, identify weaknesses in critical infrastructure, or otherwise be used in ways that are difficult to monitor once released broadly.
That anxiety has intensified across Washington in recent months as researchers, lawmakers and national-security officials debate whether the government should have more direct oversight of frontier AI models before they are publicly deployed. Some advocates argue that voluntary commitments from companies are not enough, especially as model capability advances faster than regulatory frameworks.
At the same time, industry leaders have warned that mandatory pre-release review could become a bottleneck, particularly if the government effectively gains power to delay launches or signal approval before products go to market. Critics say that could slow innovation and advantage foreign rivals.
Balancing innovation and oversight
Trump’s decision reflects a familiar split within the broader AI policy debate. On one side are those who want stronger federal oversight to reduce risks tied to misinformation, cybersecurity and autonomous behavior in increasingly capable systems. On the other are those who fear that heavy-handed regulation could undercut U.S. competitiveness just as the AI sector becomes central to economic growth, defense and industrial strategy.
The administration has consistently emphasized “global dominance” and American leadership in AI. That message has shaped a broader approach that favors deregulation, infrastructure expansion and a lighter-touch framework for the industry. Trump has repeatedly signaled that he sees AI as a strategic race that the United States must win, especially as China expands investment in the sector.
The shelved order would have represented a more interventionist step than some in the White House were comfortable with. Requiring early model access for government review could have marked a shift toward a more formal approval process, even if the policy stopped short of outright licensing.
What tech companies were watching
Tech firms and AI developers were closely watching the proposal because it could have changed the release process for cutting-edge models. Companies would likely have faced added compliance obligations and more direct scrutiny over model safety testing, cybersecurity and potential misuse scenarios.
Supporters of stronger oversight argue that such a process may become unavoidable as AI systems grow more capable. They contend that companies developing frontier models should not be the sole judges of when a system is safe enough to deploy.
But the industry also fears that a pre-release government review regime could create uncertainty around launch timelines, expose sensitive intellectual property, and open the door to political or bureaucratic delays. Those concerns appear to have weighed heavily in the administration’s decision-making, especially given Trump’s own emphasis on speed, competition and national advantage.
A broader AI policy fight
The episode does not signal a retreat from AI policymaking, but rather a shift in emphasis. Under Trump, the White House has focused heavily on removing barriers to AI development while resisting efforts that could be interpreted as slowing the sector down.
That approach has already shaped other AI actions, including efforts to limit what the administration sees as “woke” or ideologically biased AI in government use and broader moves to favor a national policy framework over a patchwork of state rules. The administration has repeatedly argued that the United States needs a unified strategy to maintain technological leadership.
Still, the abandoned order highlights the difficulty of writing AI rules that satisfy both security hawks and innovation advocates. Even within the same administration, those goals can point in opposite directions. One side wants visibility, controls and safety checks; the other wants speed, scale and fewer obstacles.
For now, Trump’s reversal leaves the White House without the new AI order it had been preparing to unveil. But the debate behind the decision is far from over. As AI models become more powerful and more widely used, pressure is likely to build for some kind of federal framework that addresses security concerns without, in the president’s words, getting in the way of America’s lead.
Reporting note: The White House has not publicly released the shelved order, and details of the draft remain limited to accounts from reporting and broadcast coverage. The broader policy debate, however, is now firmly on the national stage.