Musk, Zuckerberg and Sacks Persuade Trump to Drop AI Order

President Donald Trump canceled a planned executive order on AI after direct conversations with Elon Musk, Mark Zuckerberg and adviser David Sacks between Wednesday night and Thursday morning. The administration official reason for the withdrawal was to protect the United States’ competitive position versus China.

The unsigned order would have created a voluntary process for developers to share advanced models with federal agencies for security review up to 90 days before public release. It did not include mandatory licensing or compulsory hold periods. The proposal had been delayed multiple times before the decision not to sign.

In the Oval Office, Trump told reporters he worried the measure “could have been a blocker” and said he did not want anything that would get in the way of U.S. leadership in AI. He declined to identify the specific provisions he found objectionable.

Musk’s xAI operates in the same market as several leading model makers. Meta has promoted open-source approaches to AI development. David Sacks left a formal White House advisory role in March but continued to speak with senior aides and took part in discussions with the president.

Separate federal activity is proceeding. The Centre for AI Standards and Innovation has signed evaluation agreements that allow government assessors to review models from major companies before they become publicly available. Those agreements remain in effect independent of the unsigned executive order.

Congress has not passed comprehensive AI legislation. Federal policy to date has been built through executive actions, agency guidance and voluntary industry arrangements. The White House previously urged Congress to create a single national standard to avoid a patchwork of state laws; lawmakers have not enacted such a standard.

Chinese authorities have accelerated AI rulemaking, including plans for broader legislation and requirements for companies to form internal ethics review committees. Lizzi C. Lee of the Asia Society Policy Institute noted governments are debating how to govern powerful AI without blocking innovation. Kyle Chan of the Brookings Institution added that regulation and safety measures can be designed to protect both security and innovation.

With the order unsigned, voluntary programs and agency-led reviews remain the primary federal paths for AI oversight while lawmakers and regulators consider next steps.