Trump abruptly cancels AI safety executive order following pressure from tech billionaires

President Donald Trump pulled a highly anticipated executive order on artificial intelligence safety at the eleventh hour, following a series of urgent phone calls from prominent Silicon Valley leaders[1]. The sudden cancellation occurred just hours before a scheduled White House signing ceremony, catching many industry executives by surprise as they were already traveling to Washington to attend[1]. The tech heavyweights who intervened included SpaceX and Tesla Chief Executive Officer Elon Musk, Meta founder Mark Zuckerberg, and venture capitalist David Sacks, who previously served as the administration’s informal artificial intelligence and cryptocurrency adviser[1][2]. In the Oval Office, Trump explained his abrupt decision to reporters, stating that he did not want to enact any policy that could act as a blocker to American technological progress[1]. He emphasized that the United States maintains a tenuous lead over global competitors like China, and declared his refusal to allow regulatory frameworks to hinder that advantage[3][4]. This dramatic reversal highlights the extraordinary influence that a select group of tech billionaires wields over the administration's policy direction, leaving the future of national AI governance in a state of deep uncertainty[1].

The aborted executive order was designed to establish a voluntary verification and review system aimed at evaluating the potential national security and cybersecurity risks of the world’s most advanced artificial intelligence systems[3][5]. Under the draft proposal, participating technology companies would have voluntarily shared information about their upcoming frontier AI models with federal authorities up to ninety days prior to their public release[2][6]. This pre-launch window was intended to give government experts and cybersecurity officials an opportunity to pressure-test the systems for severe hazards, such as the ability to assist in cyberwarfare or compromise critical infrastructure[3][6]. Originally conceived as a more rigid oversight program, the policy had already been significantly watered down to a purely voluntary framework in an effort to accommodate the concerns of major technology developers[5]. Despite these concessions, the proposed ninety-day vetting period ultimately proved unacceptable to key industry leaders who argued that even voluntary government oversight would introduce bureaucratic inertia into a hyper-competitive field[1].

The successful lobbying effort to derail the executive order was orchestrated by an accelerationist faction of tech executives who argue that speed is the ultimate determinant of national security[4]. During their last-minute phone calls with the president, Musk, Zuckerberg, and Sacks warned that any regulatory hurdles, however well-intentioned, would slow down the development of a technology that is increasingly vital to the American economy[1][5]. They appealed directly to key administration officials within the National Economic Council and the vice president’s office, arguing that the true risk to the nation lies not in AI development itself, but in falling behind foreign adversaries[4]. China is actively working to eliminate its reliance on Western semiconductor technology and catch up to the United States in the field of artificial intelligence[4]. Tech leaders argued that a ninety-day delay in deploying new models would hand a significant strategic advantage to Chinese competitors, who operate without similar regulatory constraints[7][4]. The argument resonated deeply with Trump, whose broader economic and foreign policy platforms are heavily focused on preserving American manufacturing and technological hegemony[1][3].

The collapse of the executive order occurs against a backdrop of mounting alarm from national security officials and financial regulators regarding the weaponization of next-generation AI[3]. Public safety advocates and government experts have warned that advanced models are growing so sophisticated that they pose existential risks to public infrastructure and global financial systems[4]. A major catalyst for the drafted order was the emergence of highly advanced new models, such as Anthropic’s frontier model, code-named Claude Mythos[3][5]. This specific model demonstrated an unprecedented capacity to rapidly discover and exploit complex cybersecurity vulnerabilities in software code at a speed and scale far exceeding human capabilities[3][5]. The breakthrough prompted Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell to convene an urgent, closed-door meeting with Wall Street chief executives[3]. During that briefing, officials warned financial institutions of the immediate cyber risks posed by such powerful technologies and discussed the necessity of establishing best practices[3]. Proponents of the executive order argued that without some level of government oversight, state-sponsored hackers and hostile actors could easily exploit these commercial models to cripple American networks[5].

The decision to scrap the federal directive has exposed deep divisions within the artificial intelligence sector and shifted the battleground for regulation to the state level[4]. While prominent accelerationists celebrated the demise of the order, other major industry players, including OpenAI, had expressed support for the federal framework as a way to build public trust and establish clear safety baselines[4]. In the wake of the White House’s retreat, OpenAI is reportedly pivoting its regulatory advocacy efforts toward individual state legislatures[4]. The company has received a quiet green light from the White House to pursue this state-level strategy, representing a major shift from the administration's previous threats to penalize states that enact independent AI restrictions[4]. Meanwhile, local governments are already moving to fill the regulatory vacuum left by Washington[4]. In California, Governor Gavin Newsom recently signed his own executive order addressing the disruptive effects of artificial intelligence, focusing primarily on workforce retraining, severance policies, and unemployment protections in response to massive tech sector layoffs[8]. The stark contrast between California’s proactive worker protections and the federal government’s deregulation reflects a highly fragmented landscape for the future of AI policy[8][9].

By walking away from the executive order, the administration has reaffirmed a staunchly hands-off, market-driven approach to the artificial intelligence boom[5]. The decision signals to the global tech industry that the United States intends to prioritize absolute development speed and market dominance over precautionary safety measures[7]. While this stance may appease Silicon Valley investors and fuel a surge of domestic innovation, it leaves the federal government with virtually no formal mechanism to monitor or mitigate the severe cybersecurity and national security threats posed by frontier models[3][5]. As commercial entities continue to release increasingly autonomous and powerful systems into the wild, the lack of centralized oversight means that the responsibility of policing these dual-use technologies will fall entirely on the very corporations that profit from them.