White House scraps landmark AI safety order after intense lobbying from tech billionaires

A highly anticipated White House ceremony was fully scheduled, and invitations had already been dispatched to the chief executive officers of America’s leading technology companies[1][2]. Some of the industry’s most prominent figures were already in transit to Washington when the entire event was abruptly called off[3]. In a sudden reversal of policy, the administration halted the signing of a landmark executive order on artificial intelligence and cybersecurity[4]. The decision to scrap the directive came after an intense, eleventh-hour lobbying effort from some of Silicon Valley’s most influential billionaires, who successfully argued that the proposed government oversight would cripple domestic innovation and hand a decisive geopolitical advantage to global competitors, most notably China[5][2].

The decision marks a dramatic victory for the accelerationist faction of the tech industry, which advocates for rapid development with minimal regulatory friction[6]. Hours before the scheduled signing, influential figures—including the founder of xAI Elon Musk, the chief executive of Meta Mark Zuckerberg, and a key venture capitalist David Sacks who recently served as a special adviser on artificial intelligence and cryptocurrency—made direct appeals to the president[7][2]. They warned that even the relatively modest verification and review systems proposed in the draft would act as a bottleneck, slowing down the release of advanced software at a critical moment when the nation is locked in a tight technological race with Beijing[5][1].

Responding to these warnings, the president announced the postponement of the executive order directly from the Oval Office, stating that he disliked certain aspects of the draft[7][8]. In remarks to reporters, the president noted that he believed the proposed rules could get in the way of the nation's progress, emphasizing that the country is currently leading China and everyone else, and that he did not want to do anything to compromise that lead[7]. He explicitly stated that he thought the policy could have been a blocker, adding that he wanted to make sure that did not happen[1][9]. This reasoning aligned perfectly with the arguments put forward by tech leaders who asserted that the nation’s economic and national security are best served by allowing local tech companies to operate without federal guardrails[1][3].

Had it been signed, the scrapped executive order would have established a new, albeit voluntary, oversight framework for the evaluation of highly capable frontier models[7][2]. Under the proposed system, artificial intelligence laboratories would have been encouraged to share their most powerful models with federal agencies before releasing them to the public[8][10]. The draft outlined a process wherein agencies like the National Security Agency would conduct rigorous safety testing, particularly focusing on the model's cybersecurity capabilities, within a specified pre-release window[2][11]. The Treasury Department was slated to play a central role as well, serving as a hub to identify vulnerabilities in the models and distribute fixes to protect critical infrastructure, including hospitals and financial institutions[7][9].

The drive to draft this oversight framework was largely fueled by rising anxieties within the national security establishment over a new class of highly autonomous, cyber-focused models[12][11]. Intelligence agencies have voiced growing alarm over systems capable of executing complex tasks over extended periods, particularly those demonstrating a striking ability to identify and exploit software vulnerabilities[1][13]. For instance, the recent preview release of a highly advanced model, Claude Mythos, by the safety-focused laboratory Anthropic, sent shockwaves through the cybersecurity sector[7][14][15]. This model demonstrated the ability to operate autonomously for hours to discover zero-day vulnerabilities in real, open-source code bases—capabilities that national security experts fear could easily be weaponized by hostile nation-states if left entirely unchecked[13][14].

This tension has exposed deep divisions within Silicon Valley itself[1]. While some firms strongly opposed the executive order, others, including the creators of some of the world's most popular generative systems, actively supported the federal framework[6]. Larger, well-capitalized firms often favor uniform federal rules as a way to establish regulatory clarity and prevent a chaotic patchwork of differing state-level laws[16][17]. Following the cancellation of the executive order, some of these safety-conscious companies have reportedly shifted their focus toward enacting regulations at the state level, a strategy that the White House has reportedly given a quiet green light to pursue despite earlier threats to challenge states that pass unwelcome technology restrictions[6].

The sudden abandonment of the order also highlights the profound policy influence wielded by specific industry insiders who have championed a hands-off philosophy[3][18]. Proponents of this hands-off approach argue that the domestic technology sector is already facing significant structural headwinds, including domestic energy shortages and supply chain constraints that complicate massive data center buildouts[6]. Adding regulatory delays, they argue, would only compound these issues[6]. This viewpoint is consistent with previous efforts by the administration, which recently issued an executive action aimed at challenging restrictive state-level regulations through legal action and funding cuts in an effort to enforce a minimally burdensome national framework[2][19].

Yet, the decision to walk away from federal oversight leaves a policy vacuum at a time of increasing public and political friction[10]. While tech billionaires have successfully captured the ear of the administration, the broader electorate is expressing growing anxiety over the unchecked march of artificial intelligence[1][10]. Public concerns continue to mount regarding potential widespread job displacement, spiraling energy costs driven by massive data centers, and the real-world cybersecurity threats posed by highly autonomous digital agents[6][1][10]. The cancellation of the order represents a high-stakes gamble that prioritizing raw technological speed over safety testing will ultimately yield the safest and most prosperous outcome for the nation[5].

Ultimately, the scrapping of this executive order cements a highly deregulatory path for the domestic artificial intelligence sector, leaving its future development almost entirely in the hands of private corporations[20][10]. Meanwhile, global competitors are taking a starkly different approach[21]. Across the Pacific, rival governments are accelerating comprehensive legislative work plans to establish rigid state-controlled governance structures for their own domestic technology sectors[21]. By choosing to step back from federal oversight, the administration has signaled its belief that private-sector speed is the ultimate weapon in the global technological cold war, even as the line between commercial innovation and national security becomes increasingly blurred[5][11].