The abrupt collapse of President Donald Trump's planned executive order on artificial intelligence has sent shockwaves through Washington and Silicon Valley, with conflicting accounts over who was responsible. The order, which would have required developers of powerful AI models to give the federal government up to 90 days of pre-release access for security testing, was pulled by Trump on Thursday after weeks of internal preparation. A leaked draft revealed the White House had prepared new cybersecurity measures targeted at AI-enabled threats to critical infrastructure, including banks, hospitals, and government systems.
Speculation immediately centered on Trump's influential allies in the tech industry, particularly those who have long advocated for minimal government oversight of AI development. According to reports from Politico and other outlets, David Sacks, the Silicon Valley venture capitalist who served as Trump's AI and crypto czar, made a direct call to the president Thursday morning — blindsiding White House staff — to warn that the order would stifle innovation and harm the United States' competitive edge in the global AI race against China.
The Washington Post later expanded the narrative, reporting that last-minute calls from Sacks, SpaceX and Tesla CEO Elon Musk, and Meta CEO Mark Zuckerberg collectively convinced Trump to scrap the order. However, both Musk and Meta quickly disputed the claim. Musk posted on his social media platform X: "This is false. I still don't know what was in that executive order and the president only spoke to me after declining to sign." Meta also released a statement asserting that Zuckerberg had spoken to Trump only after the order was rescinded, denying any influence over the decision.
Background of the Executive Order
The shelved order was triggered by growing concerns over the capabilities of advanced AI models, particularly Anthropic's "Mythos" model, which the startup refused to release publicly due to its ability to exploit vulnerabilities in computer systems — including those of banks, governments, and hospitals. The order sought to establish a structured process for federal oversight, allowing the government to assess safety risks before such models reached the public domain.
Under the draft, companies developing AI models exceeding certain computational thresholds would voluntarily submit them to the National Institute of Standards and Technology (NIST) for evaluation. The government would then have up to 90 days to analyze the model and issue recommendations, after which the company could proceed with public release. Importantly, the draft explicitly stated that nothing in the order should be interpreted as creating a mandatory licensing or approval requirement — a concession aimed at allaying fears of government overreach.
Despite this safeguard, opposition from key tech figures was strong. Sacks, speaking on his "All-In" podcast earlier this month, argued that pre-release government approvals were "solving a problem that didn't really exist" because companies like Anthropic were already voluntarily keeping dangerous models from the public. He accused what he called "AI ideologues or doomers" of using the Mythos controversy to "create a permanent new infrastructure in Washington."
The Role of Silicon Valley in AI Policy
The episode highlights the deepening rift between two factions within the US tech landscape. On one side are advocates for safety and regulation, including many researchers and civil society groups, who argue that powerful AI systems pose existential risks if left unchecked. On the other side are industry leaders and investors who fear that any form of government oversight will slow innovation, cede leadership to China, and create unnecessary bureaucracy.
Elon Musk, despite his public stances on AI risk — he co-founded OpenAI with the mission of safe AI development and has called for a moratorium on training models more powerful than GPT-4 — has also been a vocal critic of excessive regulation. His denial of involvement in the order's collapse suggests he may be distancing himself from the political maneuvering, even as his companies, including xAI, compete in the AI space.
David Sacks, a close Trump ally and venture capitalist at Craft Ventures, has been a central figure in shaping the administration's tech policy. His position as AI czar gave him direct access to the president, and his opposition to the order likely carried significant weight. Reports indicate that Sacks had earlier expressed support for the order but changed his mind after reviewing the draft, fearing it could evolve into a mandatory regulatory framework.
Mark Zuckerberg's involvement, though disputed, reflects Meta's strategic interest in avoiding government oversight. Meta has invested heavily in AI research and deployment, including large language models like Llama, which are released openly. Any requirement for pre-release government access could complicate Meta's development cycle and global rollout.
Comparison with International Approaches
The collapse of the order leaves the United States without a formal federal strategy for managing AI security risks, a stark contrast to the European Union's comprehensive AI Act, which entered into force in 2024. The EU regulation sets binding rules for high-risk AI systems, including mandatory transparency requirements, risk assessments, and incident reporting. For the most powerful general-purpose AI models, the EU imposes obligations on safety testing, documentation, and cooperation with regulators.
In Asia, China has implemented stringent controls on AI, including requiring companies to obtain licenses before releasing generative AI systems to the public. Japan and South Korea are developing their own regulatory frameworks, while the United Kingdom has taken a more permissive approach under the banner of "pro-innovation" regulation.
Trump had previously rescinded an AI oversight executive order signed by his predecessor Joe Biden in 2023, which required AI companies to share safety test results with the government and relied on voluntary commitments. Biden's order was already considered a light-touch approach by many experts, who deemed it insufficient to address the accelerating pace of AI capabilities.
Implications for the Future
With no executive order in place, the Trump administration currently lacks any formal mechanism to assess or mitigate risks from the most advanced AI models. The Department of Homeland Security and other agencies may continue ad-hoc efforts, but without presidential backing, those initiatives lack funding and authority.
The episode also underscores the influence of a small circle of tech billionaires on national policy. The ability of Sacks, Musk, and Zuckerberg — whether directly or indirectly — to derail a carefully crafted executive order raises questions about democratic accountability and the role of corporate power in shaping AI governance.
Proponents of regulation argue that voluntary measures are insufficient, as companies have financial incentives to prioritize speed over caution. The Mythos model, which prompted the order, is a case in point: Anthropic itself refused to release it, but other firms might not be as conscientious. The European approach, with legally binding requirements, is seen by many as the gold standard for ensuring safety without stifling innovation.
Conversely, the tech industry warns that government pre-release access could become a backdoor to censorship or industrial espionage, especially if the review process is politicized. The leaked draft's explicit statement against mandatory licensing was intended to counter such fears, but skeptics remain unconvinced.
The future of AI policy in the US remains uncertain. Congress has yet to pass comprehensive AI legislation, and partisan divides on the issue are deep. Some Republicans favor minimal regulation, while some Democrats push for aggressive safeguards. The White House may attempt to craft a narrower order that addresses cybersecurity without triggering opposition from Silicon Valley, but the current failure suggests a lack of consensus even within the administration.
In the meantime, AI companies continue to develop and deploy models at a rapid pace. The absence of federal oversight leaves a vacuum that states like California are beginning to fill. California's proposed AI safety bill, which would require companies to test and report on large models, has stirred intense debate and could set a precedent for other states, potentially creating a patchwork of regulations that the tech industry finds onerous.
The collapse of Trump's executive order is not the end of the story — it is a chapter in an ongoing struggle between innovation and safety, between industry influence and public interest. As AI capabilities grow exponentially, the need for thoughtful governance becomes more urgent. Whether the United States will rise to the challenge or continue to cede leadership to Europe and Asia remains an open question.
Source: MSN News