Trump Halts AI Executive Order, Regulatory Concerns Trump Competition Anxiety
Original Title: Donald Trump abruptly postpones AI order after White House infighting
Original Author: Joe Miller, Financial Times
Translation: Peggy
Editor's Note: The Trump administration had initially sought to establish a pre-release security testing mechanism for cutting-edge AI models, but hours before signing, this executive order was abruptly put on hold.
According to the original plan, top AI companies such as OpenAI, Google, Anthropic, and xAI were to voluntarily share their models with the U.S. government for national security and cybersecurity risk assessments 90 days before the public release of advanced models. Trump ultimately refused to approve, citing that he "did not want anything to hinder America's leadership position in AI."
Behind this last-minute reversal lies a core contradiction facing U.S. AI policy: the capabilities of cutting-edge models have begun to touch on public risks such as cybersecurity vulnerabilities, financial system vulnerabilities, job displacement, and data center expansion. However, against the backdrop of the U.S.-China AI competition being seen as a national strategic issue, any regulatory arrangement could be interpreted by the industry as slowing down innovation and weakening U.S. competitiveness.
More subtly, this executive order was originally not a mandatory approval system, but a "collaborative, voluntary" model evaluation framework. In other words, the White House was not attempting to directly control model releases but was hoping to add a government security testing step before the models were made available to the public. Yet even this relatively mild approach was temporarily put on hold amid the tug-of-war between security governance and technological advancement.
AI is transitioning from a simple technology industry issue to a crossroads of national security, macroeconomics, and political governance. The focal point of the U.S. AI policy debate is shifting from "whether to support AI development" to "how to manage the systemic risks that cutting-edge models may bring without sacrificing our leading edge."
The following is the original text:
Donald Trump's sudden reversal came after weeks of internal U.S. government disputes over the appropriate scope of regulatory oversight.
The White House unexpectedly delayed the long-awaited signing of an artificial intelligence executive order. Previously, Donald Trump stated that he "did not like" certain parts of the plan, especially the U.S. government's proposed arrangement to conduct national security and cyber risk reviews of AI models.
The executive order was originally scheduled to be signed on Thursday afternoon. As planned, top AI companies such as OpenAI, Google, and Anthropic have voluntarily committed to submit their models to the government for review.
Trump's sudden reversal came after weeks of internal debate within the U.S. government over the boundaries of regulatory review.
Referencing the executive order, Trump stated, "I don't like certain aspects of it. We are ahead of China, and we are ahead of everybody, and I don't want anything to slow us down in our quest for maintaining that lead." He also mentioned that artificial intelligence is "bringing in a lot of jobs."
Prior to the abrupt postponement of the signing ceremony, several tech CEOs had planned to travel to Washington to attend the event with Trump.
U.S. President Donald Trump, concerned about U.S. innovators falling behind in competition with China, refused to approve this AI executive order.
The delayed signing comes as multiple polls continue to show that American voters are concerned about the impact of artificial intelligence, with many supporting stricter regulations and safety guardrails for this emerging technology.
Public concerns regarding the security implications of open-sourcing powerful AI models are on the rise; meanwhile, controversies surrounding the impact of AI on employment and the construction of large-scale data centers have added complexity to the White House's political calculus. Just recently, this administration's stance towards the AI industry was notably positive.
Some of Trump's allies had called for leading AI models to be brought under U.S. government control; however, other MAGA supporters warned that any measures restricting AI growth could potentially weigh down the U.S. economy.
A recent poll conducted for the Institute for Family Studies this month showed that 82% of Americans support the White House conducting security testing on advanced AI models.
The preparation of this executive order stemmed from key White House officials who had early access to Anthropic's latest Mythos model, including Treasury Secretary Scott Bennett. The model boasts advanced capabilities for discovering cybersecurity vulnerabilities. Informed officials stated that the issues like banking system vulnerabilities exposed by this model had left them shocked.
So far, Anthropic has only made Mythos available to a select few trusted institutions, including tech companies and some banks, to allow these entities to proactively discover and address cybersecurity issues before hackers gain access to the model.
Former President Trump's National Economic Council director, Kevin Hassett, once proposed that cutting-edge AI models should undergo a similar approval process as drugs, only being officially released once "proven safe, just like the FDA approves drugs."
This statement faced strong opposition from AI founders and investors, including some closely tied to the Trump administration, who argued that such a regime would weaken America's innovation prowess.
The executive order falls far short of the demands of a similar approval regime, instead moving to establish a "collaborative, voluntary model benchmarking framework." By Thursday morning, before the scheduled signing ceremony, White House officials had briefed reporters on the contents of this executive order.
Under the proposed agreement, top AI companies, including OpenAI and xAI, will voluntarily share models with the government 90 days before the models are released to the public. In essence, this mechanism will still heavily rely on the willingness of AI company leaders to cooperate.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
