President Donald Trump abruptly hit pause on a long-teased executive order on artificial intelligence on Thursday, shelving a White House signing ceremony just hours before it was set to begin and leaving a roster of invited tech and finance heavyweights scrambling. Sitting in the Oval Office, Trump told reporters he had problems with parts of the draft and worried it could slow the United States in the AI race. "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," he said.
As reported by The Associated Press, the administration said the signing was called off after the president reviewed the order's text and that the event will be rescheduled. Axios reported that the White House had invited major tech, AI and cybersecurity executives to the ceremony, highlighting the push to win industry support for whatever new oversight rules eventually emerge.
Why Officials Sounded the Alarm
The delay lands as senior officials grow more vocal about AI's cybersecurity risks following an urgent April meeting at Treasury headquarters, where Treasury Secretary Scott Bessent and Fed Chair Jerome Powell convened leaders of major banks. Bloomberg reported that officials warned bank CEOs the Anthropic model known as Claude Mythos could accelerate the discovery of software vulnerabilities and urged institutions to ramp up their own defensive testing.
Anthropic's Mythos and Project Glasswing
Anthropic has said in its system card and transparency hub that Claude Mythos Preview showed unexpectedly powerful coding and red-teaming abilities in internal tests, including the capacity to autonomously identify zero-day flaws, and that the company restricted access to the model to a small group of partners for defensive work under "Project Glasswing." Anthropic's disclosures have prompted regulators and firms to treat some frontier models as tools that can both bolster security and open up new attack surfaces.
Tension Between Safety and Competitiveness
Inside the White House, officials have been trying to balance a more precautionary posture with the president's stated reluctance to impose rules that could crimp U.S. competitiveness. The Associated Press noted that any moves resembling government screening of commercial AI models would mark a sharp break from earlier promises to roll back Biden-era AI regulations. Axios reported internal disagreements over how strict any screening should be, a split that officials say helps explain the last-minute reversal on the signing.
The White House has not released the final text of the order and has said only that the event will be rescheduled. Spectrum Local News first posted the local bureau's version of the Associated Press report on the delay. For now, industry and regulators say they will continue with defensive testing and information-sharing, but the postponed signing leaves open when, or whether, the administration will move ahead with formal screening rules for frontier AI systems.
