The complete cancellation of President Donald Trump’s proposed artificial intelligence executive order was the result of a coordinated lobbying campaign by some of the most powerful figures in the U.S. technology industry, according to a detailed investigation by The Guardian. According to the report, Elon Musk, Mark Zuckerberg, and former White House AI czar David Sacks personally intervened to kill the order’s safety review provisions, deploying the China threat narrative as their primary lever of influence. The outcome marks a significant defeat for AI safety advocates and raises fundamental questions about the future of AI governance in the United States.
From Postponement to Cancellation
The story of this executive order’s demise began on May 21, when the White House abruptly postponed a scheduled signing ceremony. As EastFrontier reported at the time, the stated reason for the postponement was that the administration did not want to implement any measures that might slow American AI developers in the race against China. What was not publicly disclosed at the time was that the postponement was the result of direct lobbying by Musk, Zuckerberg, and Sacks, who had engaged in private conversations with President Trump in the days before the planned signing.
The Guardian’s reporting reveals that the lobbying campaign was more coordinated and more effective than the initial postponement suggested. Within 48 hours of the postponement, the safety review provisions at the heart of the order had been stripped entirely. The final version of any executive order on AI, if one is signed at all, will contain none of the mandatory pre-deployment review mechanisms proposed in the original draft.
The Architecture of the Lobbying Campaign
The proposed executive order had been in development for months and was designed to establish a voluntary government review process for new AI models on national security grounds. The review would have required AI developers to notify the government before releasing models above a certain capability threshold, giving national security agencies an opportunity to assess potential risks.
This modest proposal, voluntary, not mandatory, was nonetheless characterized by its opponents as a potentially catastrophic regulatory burden. Musk, Zuckerberg, and Sacks argued in their conversations with Trump that any form of government review, even a voluntary one, would create uncertainty, slow development timelines, and signal to the world that the U.S. was imposing constraints on its own AI industry. They framed the choice as binary: either American AI companies operate without constraint, or China wins.
The China Argument as a Policy Tool
The use of the China threat to deflect domestic AI regulation has become a recurring pattern in U.S. tech policy debates. The argument is structurally powerful: it frames safety and oversight as luxuries that a country in a competitive race cannot afford, and it positions any regulatory burden as a unilateral disarmament. As EastFrontier has documented in its coverage of the US-China AI safety dialogue, the irony is that both the U.S. and China have expressed interest in bilateral AI safety cooperation — suggesting that safety and competitiveness are not necessarily in conflict.
The Guardian’s reporting suggests that the China argument was deployed not because the lobbyists genuinely believed the order would hand China an advantage, but because it was the most effective framing for an administration that has made technological competition with China the centerpiece of its technology policy. By invoking China, the tech billionaires were speaking the administration’s language.
Implications for AI Governance
The cancellation of the safety review provisions leaves the United States without any formal mechanism for government oversight of frontier AI model deployments. The responsibility for safety rests entirely with the companies developing the models, the same companies that lobbied to kill the review process. AI safety advocates have warned that commercial incentives to rush products to market often outweigh safety considerations and that the absence of any external check creates systemic risk.
The episode also illustrates the structural challenge of governing transformative technologies when the companies developing them have both the resources and the political access to shape the regulatory environment. The three individuals who the Guardian argues effectively killed this executive order (Musk, Zuckerberg, and Sacks) collectively control companies that are among the most powerful AI developers in the world. Their ability to influence U.S. AI policy through direct access to the president is a feature of the current political moment that will shape the trajectory of AI governance for years to come.
For China, the cancellation of the U.S. AI safety review provisions is a complex development. On one hand, it removes a potential source of regulatory friction that could have slowed American AI development. On the other hand, it undermines the credibility of U.S. calls for international AI safety cooperation, including the bilateral US-China AI safety dialogue that both governments have recently agreed to pursue, as EastFrontier reported. If the United States cannot maintain even a voluntary domestic safety review process, its ability to negotiate meaningful international AI safety agreements, including with China, is significantly weakened. The episode thus has implications not just for domestic U.S. AI governance but for the global architecture of AI safety that both governments have said they want to build.