AI regulation splits Trump administration as frontier model order stalls

Trump postponed signing an executive order establishing a voluntary AI model review process hours before the planned ceremony, citing dislike of certain aspects.

Objective Facts

President Trump abruptly postponed signing an executive order on Thursday that would have established a voluntary process for AI companies to share advanced models with the government before public release, saying he "didn't like certain aspects" of the order. The framework under discussion would have given federal agencies up to 90 days for security review. Eleventh-hour phone calls from former Trump AI czar David Sacks, Meta CEO Mark Zuckerberg, and xAI CEO Elon Musk between Wednesday night and Thursday morning persuaded Trump to shelve the order. Trump "just hates regulation," according to sources, with Sacks also opposing the order as "unnecessary" and "just something doomers wanted". Trump told reporters he did not want to do anything to get in the way of the U.S. lead over China in AI.

Left-Leaning Perspective

NPR's Deepa Shivaram reported on the alarm triggered by Anthropic's Mythos model and noted that "officials started to talk about AI and use words like safety for the first time," with Trump even saying regulations around AI should exist in a Fox News interview. Analysis from Let's Data Science observed that Anthropic's caution with Mythos and the on-again, off-again CAISI deals became the de facto policy, depending "entirely on companies choosing restraint," working only because the most dangerous models remained mostly unreleased and described as "a thin thread to hang national cybersecurity on". Andrew Lokay, senior research analyst at Beacon Policy Advisors, told The Hill that Trump's decision "is a sign of the continuing influence of Silicon Valley with the Trump administration" and that Sacks "seems to have the ear of the president on this topic," leaving him questioning what AI regulations may come from the White House. The Hill's analysis noted the apparent divisions within the Trump administration "could inhibit future policy development," leaving states like California to fill regulatory gaps, as Gov. Newsom signed an executive order addressing AI-related job losses the same day the federal order was cancelled. Let's Data Science's analysis underscored what the left sees as a major failure: Trump left unanswered the question of what happens when a frontier model becomes the threat the order was written to address, with no order in place. The left's coverage emphasizes that the order was voluntary and minimal—not aggressive regulation—yet still couldn't survive Silicon Valley pressure, raising concerns about the government's capacity to respond to AI security threats.

Right-Leaning Perspective

Gizmodo reported that David Sacks brought up industry concerns directly to Trump, arguing that AI companies are already cooperating with the government, that any federal review process could slow innovation and give China an edge in the AI arms race, and he worried any voluntary framework would someday transform into mandatory government review. Analysis from ICO Optics noted that Sacks argued government review would "slow innovation and hurt America's edge over China" and contended that "companies already cooperate, so more government checks could actually hold things back instead of helping the public". RT Business reported the order would have marked a major shift from Trump's hands-off AI policy, noting that since returning to office he has "championed minimal regulation, pledged to make the US an 'AI state,' and called the technology 'a beautiful newborn baby' vital for competing with China". AI News noted that "in the current administration, the effective veto on AI regulation sits with a small group of industry principals who have direct access to the president," with Musk having "a structural interest in keeping the regulatory field open" as xAI competes with OpenAI and Anthropic. Dean Ball, one of the primary authors of the White House AI Action Plan, wrote on social media that "If you needed yet more evidence that the burden of frontier AI governance is going to rest principally on the private sector, you got it yesterday," framing the cancellation as validation of the accelerationist position. Right-leaning coverage largely frames this as Trump correctly prioritizing competitiveness over regulatory burden, with the order's cancellation as appropriate deference to industry expertise on innovation timelines.

Deep Dive

The specific angle of this story is a rare moment of internal Trump administration fracture over AI regulation, broken in real-time by a single entrepreneur's phone call. White House officials believed David Sacks was "generally happy" with the order through the week until he raised concerns Wednesday night and called Trump Thursday morning; the order was "not killed on substance over weeks of negotiation; it was killed in 12 hours by a single phone call". The deeper story is about how AI policy uncertainty emerges not from deliberation but from proximity to power: "In the current administration, the effective veto on AI regulation sits with a small group of industry principals who have direct access to the president," with Musk having "a structural interest in keeping the regulatory field open" as xAI competes with OpenAI and Anthropic. Both sides mischaracterize what they won and lost. The right argues it protected innovation by preventing a regulatory "blocker," but the federal government "drafted, staged, and abandoned a frontier-model review process in the span of a single afternoon," with "the pattern...consistent: the administration keeps building oversight machinery and keeps getting cold feet about turning it on". This suggests not victory for deregulation but incoherence. The left argues that relying on voluntary compliance is dangerous, but even the proposed order would have asked companies to "share a covered model...as much as 90 days before shipping it"; participation would have been voluntary; there was no requirement models pass review before launch; and unlike Biden's order, it would not have forced labs to hand over internal testing results. The order itself was a minimalist compromise that satisfied neither side. What happens next depends on whether frontier AI models trigger a visible security incident before the next policy attempt. "The agencies, the definitions of a 'covered model,' and the 90-day idea are all written down somewhere in a White House drawer. They will come back, probably after the next model does something frightening in public." The real test of who won this battle—industry or security advocates—will come when the costs of inaction become undeniable rather than theoretical.

OBJ SPEAKING

Create StoryTimelinesVoter ToolsRegional AnalysisPolicy GuideAll StoriesCommunity PicksUSWorldPoliticsBusinessHealthEntertainmentTechnologyAbout

AI regulation splits Trump administration as frontier model order stalls

Trump postponed signing an executive order establishing a voluntary AI model review process hours before the planned ceremony, citing dislike of certain aspects.

May 26, 2026· Updated May 28, 2026
What's Going On

President Trump abruptly postponed signing an executive order on Thursday that would have established a voluntary process for AI companies to share advanced models with the government before public release, saying he "didn't like certain aspects" of the order. The framework under discussion would have given federal agencies up to 90 days for security review. Eleventh-hour phone calls from former Trump AI czar David Sacks, Meta CEO Mark Zuckerberg, and xAI CEO Elon Musk between Wednesday night and Thursday morning persuaded Trump to shelve the order. Trump "just hates regulation," according to sources, with Sacks also opposing the order as "unnecessary" and "just something doomers wanted". Trump told reporters he did not want to do anything to get in the way of the U.S. lead over China in AI.

Left says: Without the order, national cybersecurity depends on companies voluntarily choosing restraint, described as "a thin thread to hang national cybersecurity on." The administration's divisions could inhibit future policy development, leaving states like California to fill regulatory gaps as Gov. Newsom signed his own AI executive order the same day.
Right says: Trump told reporters "we're leading China, we're leading everybody, and I didn't want to do anything to get in the way of that lead," framing the order cancellation around competitive advantage rather than substance. Sacks argued companies are already cooperating voluntarily and that any federal review process would slow innovation and could eventually become mandatory.
✓ Common Ground
Both sides acknowledge that Anthropic's Mythos model posed unprecedented cybersecurity concerns, that the company voluntarily limited its release to a select consortium, and that U.S. officials had been pushing to make Mythos more widely available to federal agencies to test networks for security flaws—with the White House subsequently rejecting Anthropic's plans to distribute it to additional companies and organizations, citing security concerns.
Both left and right outlets acknowledge a genuine conflict within Trump's own political base over AI oversight, driven by the release of powerful new AI systems including Mythos, though some cybersecurity executives dispute whether those fears are overblown.
There is apparent agreement across analyses that the proposed order was notably non-aggressive: participation in model testing would have been voluntary, not mandatory; there was no requirement that any model pass government review before launch; and unlike Biden's order, it would not have forced labs to hand over internal testing results.
Objective Deep Dive

The specific angle of this story is a rare moment of internal Trump administration fracture over AI regulation, broken in real-time by a single entrepreneur's phone call. White House officials believed David Sacks was "generally happy" with the order through the week until he raised concerns Wednesday night and called Trump Thursday morning; the order was "not killed on substance over weeks of negotiation; it was killed in 12 hours by a single phone call". The deeper story is about how AI policy uncertainty emerges not from deliberation but from proximity to power: "In the current administration, the effective veto on AI regulation sits with a small group of industry principals who have direct access to the president," with Musk having "a structural interest in keeping the regulatory field open" as xAI competes with OpenAI and Anthropic.

Both sides mischaracterize what they won and lost. The right argues it protected innovation by preventing a regulatory "blocker," but the federal government "drafted, staged, and abandoned a frontier-model review process in the span of a single afternoon," with "the pattern...consistent: the administration keeps building oversight machinery and keeps getting cold feet about turning it on". This suggests not victory for deregulation but incoherence. The left argues that relying on voluntary compliance is dangerous, but even the proposed order would have asked companies to "share a covered model...as much as 90 days before shipping it"; participation would have been voluntary; there was no requirement models pass review before launch; and unlike Biden's order, it would not have forced labs to hand over internal testing results. The order itself was a minimalist compromise that satisfied neither side.

What happens next depends on whether frontier AI models trigger a visible security incident before the next policy attempt. "The agencies, the definitions of a 'covered model,' and the 90-day idea are all written down somewhere in a White House drawer. They will come back, probably after the next model does something frightening in public." The real test of who won this battle—industry or security advocates—will come when the costs of inaction become undeniable rather than theoretical.

◈ Tone Comparison

Left-leaning coverage emphasizes alarm and systemic risk: Let's Data Science describes relying on voluntary compliance as "a thin thread to hang national cybersecurity on," while observers speak of "continuing influence of Silicon Valley." Right-leaning framing emphasizes competitive advantage and innovation: sources cite Sacks's argument about China gaining "an edge" and Trump's characterization of AI as "a beautiful newborn baby" vital to competition. The right frames the cancellation as principled deregulation; the left frames it as industry capture.