Trump postpones AI security executive order after pressure from Sacks, Musk, and Zuckerberg, despite majority American support for AI regulation

David Sacks, Elon Musk, and Mark Zuckerberg lobbied Trump on Thursday to block a voluntary AI model vetting executive order. The order would have allowed AI companies to submit advanced models to national security agencies for testing and vetting up to 90 days prior to release. Sacks, who stepped down as Trump's AI and crypto czar in late March, called Trump to voice concerns that the voluntary system could act as a de facto licensing regime.

Public opinion favors AI oversight: a poll by the Future of Life Institute found 79% of Republican voters support government testing of AI models before release, and 87% favor government power to block releases posing national security threats. More than 60 MAGA loyalists, including former Trump advisor Steve Bannon, signed an open letter urging Trump to test and approve powerful AI models before release.

The Trump administration is conducting ad hoc oversight of Anthropic's Mythos model—which possesses advanced cyber capabilities—under a program called Project Glasswing that shares the model with select U.S. technology companies and financial institutions. Meanwhile, OpenAI's GPT-5.5, which is almost as capable as Mythos according to OpenAI's testing and the U.K. government's AI Security Institute, is available through a less restricted 'trusted access' program that has not faced the same government scrutiny.