Trump Cancels AI Safety Testing Executive Order After Key Tech CEOs Declined to Attend Signing

President Donald Trump abruptly canceled a planned executive order signing event on Thursday, just hours before it was scheduled to take place, after learning that several top artificial intelligence company executives had declined to attend — underscoring deepening tensions within his administration over how aggressively to govern frontier AI models.

Trump had given executives roughly 24 hours' notice to appear at the White House for the signing. When he learned that some could not make the event, he pulled the plug. Some AI company leaders who had rearranged their schedules on short notice were reportedly midair en route to the Oval Office when the cancellation was announced.

The executive order would have authorized the federal government to test frontier AI models prior to their public release, with the stated goal of identifying cybersecurity vulnerabilities before they could be exploited against banks, utilities, and other sensitive sectors.

Publicly, Trump told reporters he had reservations about the order itself. "I really thought [the order] could have been a blocker," he said, framing mandatory government testing as a potential drag on America's lead over China in AI development. "I think it gets in the way of — you know, we're leading China, we're leading everybody, and I don't want to do anything that's going to get in the way of that lead," Trump added.

The cancellation had industry fingerprints on it. xAI founder Elon Musk and Meta CEO Mark Zuckerberg reportedly urged Trump to call off the signing, according to Semafor. OpenAI, by contrast, had supported the order. Musk denied any involvement, writing on X that the claim was "false" and that he doesn't "know what was in that EO," according to Reuters. David Sacks, Trump's former AI advisor whose special government employee designation expired in March, also reportedly joined the push to delay the signing.

At the core of the industry's resistance was a dispute over testing timelines. The government had sought authority to evaluate models up to 90 days before their public release. AI labs pushed for a much narrower window of just 14 days, according to people briefed on the order.

The proposal emerged after Anthropic flagged cybersecurity risks associated with its latest model, Mythos, which alarmed some members of the Trump administration. Officials from the Commerce Department and the Office of Science and Technology Policy have generally favored a lighter regulatory touch, while the Office of the National Cyber Director has argued that governance measures are needed now.

The cancellation also reflects a broader leadership vacuum on AI policy inside the White House. Following the end of Sacks' formal advisory role, sources described a "power vacuum" in the administration's AI leadership structure — even as Sacks reportedly continues to visit the White House on a weekly basis.

The episode arrives as the U.S. and China are separately working to coordinate on AI safety at the diplomatic level. Following Trump's meeting with Chinese President Xi Jinping in Beijing last week, the two countries agreed to launch an intergovernmental dialogue on AI to navigate shared national security risks, China's Foreign Ministry confirmed. Separately, U.S. State Department official Casey K. Mace told reporters on the sidelines of the APEC trade ministers' meeting in Suzhou that the U.S. is "very active in promoting U.S. AI options and solutions" across Asia, with American tech companies planning workshops at an APEC digital event in Chengdu in July.

Lizzi C. Lee, a fellow at the Asia Society Policy Institute's Center for China Analysis, said Trump faces the same fundamental dilemma as Beijing: how to guard against national security risks without stifling the pace of AI development. Lee noted that if safety testing is scoped narrowly to national security concerns, "it probably won't slow leading US labs much." She also pointed to what she described as "a separate, potentially more important race" — determining who can govern powerful AI without choking off innovation — in which China may hold a slight edge, given Beijing's accelerating regulatory activity.

Whether Trump reschedules the signing, and what modifications might be required to secure broader industry support, remains unclear.