The Trump Administration Is at War With Itself Over AI Regulation | WIRED

Overview

The Trump Administration Is at War With Itself Over AI Regulation

The Trump administration is navigating internal strife as officials try to figure out whether they can resurrect the executive order about AI regulation that President Donald Trump abruptly nixed last month, according to multiple people familiar with the matter.

Details

The conversations in the weeks since have been widely viewed as chaotic, by both key Silicon Valley players and administration officials. Some AI executives have privately told WIRED they are uncertain what a revised executive order might require, or whether one will end up being signed at all.

On May 21, Trump canceled a planned signing ceremony for the order just hours before it was scheduled to take place. He told reporters at the time that it could stifle competition domestically and reduce the advantage the US currently maintains over China in the AI race.

The most contentious section of the nixed executive order was a provision creating a voluntary framework in which AI labs like Open AI, Anthropic, and Google would give the White House early access to AI models ahead of their public release to evaluate cybersecurity capabilities.

At its core, the push for regulation reflects a recognition inside the White House that AI is fast becoming a national security concern, given the capabilities of Anthropic’s Mythos and Open AI’s GPT-5.5 models, which excel at finding vulnerabilities in legacy software systems. The effort suggests a change of heart for the administration, which initially eschewed attempts to regulate AI.

excel at finding vulnerabilities in legacy software systems

The draft executive order also suggested AI labs could submit models up to 90 days before public release, though several AI executives tell WIRED their companies may not be prepared to share models that far ahead of time. Some AI leaders and aides are hopeful the executive order could come back in revised form, with some of its less controversial provisions intact.

Whether the administration can resurrect an AI executive order now largely rests on the ability of top White House officials to corral competing factions, according to aides across multiple agencies involved in the process.

White House chief of staff Susie Wiles has taken charge of a group of top officials pushing for the executive order to be resurrected, which also includes treasury secretary Scott Bessent and national cyber director Sean Cairncross, a former Republican political operative, the aides say.

Bessent has emerged as a notable force in the administration on AI policy. In recent weeks, he has met with Anthropic CEO Dario Amodei and other AI executives to forge a path forward, the aides say. And he is expected to take a lead role in negotiating cross-border AI regulation with China.

Trump’s influential former AI czar David Sacks stands in opposition to Wiles. Sacks told Trump that the executive order would be too onerous and reportedly successfully implored him to call off the signing hours before it was set to take place. Politico earlier reported the Sacks-Wiles dynamic.

In a post on X last week, Sacks wrote, “President Trump understands that unnecessary regulation is the biggest threat to innovation in America. Winning the AI race means not only beating China but also clearing bureaucratic hurdles thrown up by state legislatures and woke politicians in DC.”

But perhaps the biggest hurdle in getting regulation back to the table remains Trump himself, the aides said. “Resolving the infighting only matters if it gets Trump to yes,” one administration official said on the condition of anonymity to speak freely about sensitive deliberations.

In a statement, White House spokesperson Liz Huston said the administration has been trying to figure out how best to balance AI regulation. “The President’s team is united in executing his bold agenda and maintaining this critical balance,” says Huston.

Other officials have distanced themselves from the executive order process amid the uncertainty. Commerce secretary Howard Lutnick has played a minimal role in the executive order process despite his interest, say two people familiar with the matter, in part because Lutnick already has early access to new AI models through a preexisting program.

The program, called the Center for AI Standards and Innovation, has been the main clearinghouse for the US government for testing and evaluating frontier AI models, without implementing formal preapproval requirements. Lutnick elevated the program’s profile by expanding its mandate at the start of Trump’s second term.

Similarly, the Pentagon has played a back-seat role in the efforts to craft the executive order. Undersecretary Emil Michael, a former top executive at Uber, has been more interested in making sure the Pentagon gets early access to frontier models, says a person familiar with the matter.

A senior administration official stressed that the internal dynamics have been fluid since Trump canceled the previous iteration of the executive order. “We’re back to the drawing board, so everything is still to play for,” says the official.

How to find us: Add WIRED.com to your preferred sources in Google

How the Canvas hack threatened thousands of schools

Big Story: I've covered robots for years—this one is eerily lifelike

Orbs, saucers, and flashes on the moon—here’s what’s in the UFO files

Key Takeaways

The Trump Administration Is at War With Itself Over AI Regulation
The Trump administration is navigating internal strife as officials try to figure out whether they can resurrect the executive order about AI regulation that President Donald Trump abruptly nixed last month, according to multiple people familiar with the matter
The conversations in the weeks since have been widely viewed as chaotic, by both key Silicon Valley players and administration officials
On May 21, Trump canceled a planned signing ceremony for the order just hours before it was scheduled to take place
The most contentious section of the nixed executive order was a provision creating a voluntary framework in which AI labs like Open AI, Anthropic, and Google would give the White House early access to AI models ahead of their public release to evaluate cybersecurity capabilities