Trump pulled a nearly-finished artificial intelligence executive order at the last minute, said he didn’t like it, and the media immediately declared that tech billionaires had hijacked the presidency — but the actual story is far more interesting than that.
Quick Take
- Trump postponed signing an AI executive order in early 2025, citing concern it would hurt U.S. competitiveness against China.
- His administration’s own White House orders explicitly state that sustaining “global AI dominance” is the governing policy of the United States.
- David Sacks, the administration’s former AI czar, raised industry concerns before the last-minute pullback, according to Politico reporting.
- Critics framed the episode as billionaire capture; the administration framed it as protecting America’s lead in the most consequential technology race in history.
Trump Pulled the Order and Said Exactly Why
Trump did not hide his reasoning. Standing before reporters, he said there were “certain aspects of this order that I did not like” and that signing it could get in the way of the United States leading the world in artificial intelligence. He named China directly. He said the U.S. and China are “fighting for it” while every other country is “way behind.” That is not a vague deflection — it is a stated national security and economic rationale delivered in plain English. Whether you agree with it or not, the framing was explicit from the start. [4]
The January 23, 2025 White House executive order that replaced the Biden-era framework states without ambiguity: “It is the policy of the United States to sustain and enhance America’s global AI dominance.” It directed agencies to immediately review all policies, directives, and regulations stemming from Biden’s 2023 AI order, with instructions that anything inconsistent with the new dominance-first posture be suspended, revised, or rescinded. That is not a subtle shift. It is a complete philosophical reversal of the prior administration’s safety-first governance model. [2]
The Billionaire Narrative Has a Real Evidence Problem
Politico reported that David Sacks, Trump’s former AI czar, raised concerns from tech companies before the withdrawal, and that framing spread fast. [8] Multiple social media posts declared that Elon Musk and Mark Zuckerberg personally convinced Trump to kill the order. The problem is that no call logs, readouts, or transcripts have surfaced to confirm those specific conversations drove the decision. The “billionaires stopped safety rules” headline is a compelling story. It is also, so far, a story built on anonymous sourcing and inference rather than documentary evidence. That does not make it false, but it does make it premature to treat as settled fact.
This matters because the “regulatory capture” narrative conveniently sidesteps the policy question entirely. If the order contained provisions that genuinely would have slowed compute deployment, restricted model training, or handed regulatory leverage to agencies with poor track records on technology, then the competitiveness argument deserves to be evaluated on its merits — not dismissed because wealthy people also happened to oppose the order.
The December 2025 Order Confirms This Was Never About Abandoning Governance
The administration’s December 11, 2025 national AI policy framework repeated the dominance language while also directing protections for children, safeguards against censorship, copyright respect, and community protections. [5] That is not the language of an administration that threw out every guardrail to please Silicon Valley. It is the language of an administration that decided the Biden-era framework was the wrong instrument for the moment and replaced it with one designed around American competitive advantage rather than precautionary restraint. You can disagree with that prioritization. Calling it reckless abandonment of all oversight is a different claim, and the record does not support it.
The Deeper Fight Is About Who Writes the Default Rules First
The real stakes in this episode are not about one postponed order. In fast-moving regulatory domains, executive-branch policy written early tends to calcify into permanent default rules before markets have time to self-correct. Biden’s 2023 AI order imposed detailed mandates across government departments at a moment when the technology was still taking shape. [1] Trump’s team looked at that framework and concluded it would lock in compliance costs and oversight structures that favor incumbents, slow deployment, and give China room to advance while American developers navigated bureaucratic clearance processes. That is a defensible read of the incentives, even if critics prefer a different balance.
This is absolutely mental that big tech bosses have managed to talk Trump out of an executive order that would have required additional safety and security testing for AI systems🤖 today – the same week the EU released its draft classification guidelines on high-risk AI systems… https://t.co/gkcFiXxqfN
— Mary-Ann Russon (@concertina226) May 22, 2026
The competitiveness argument is not automatically correct just because it is the administration’s stated position. But the reflexive counter-narrative — that any deregulatory move in AI is billionaire-driven recklessness — is equally lazy. The United States is in a genuine race with China over which country’s AI architecture becomes the global standard. That race has national security consequences that dwarf the question of which executives made phone calls before a single order was postponed. The policy debate deserves better than a villain narrative built on incomplete evidence.
Sources:
[1] Web – AI: Broad Biden Order Is Withdrawn, but Replacement Policies Are …
[2] Web – Removing Barriers to American Leadership in Artificial Intelligence
[4] Web – Trump administration rolls-back Biden AI executive order and …
[5] Web – Ensuring a National Policy Framework for Artificial Intelligence