Trump’s AI Safety Order Faces a Big Hurdle: His Own Budget Cuts

Donald Trump signed an executive order Tuesday aimed at voluntary safety testing for advanced AI models, but critics say the plan is more about optics than action. The order, which the White House had originally planned to unveil at a flashy event with top tech CEOs, was quietly released after Trump canceled the signing last month, citing fears it might slow down innovation.

The problem, according to security experts, is that the administration lacks the personnel to actually carry out these reviews. The Department of Government Efficiency—known as DOGE—has slashed cybersecurity teams across federal agencies, leaving the very offices that would oversee AI testing understaffed and overstretched. One former official described the situation as “trying to inspect a skyscraper with a single building inspector.”

The order itself is a watered-down version of earlier drafts. It creates a voluntary process for AI companies to work with the government on safety checks, but imposes no mandatory requirements. Trump’s statement promised the effort would “ensure that the best and most secure technology is deployed rapidly,” while also maintaining an “America First” approach to cybersecurity.

Industry insiders note that the final text differs little from the leaked draft that already sparked backlash from AI firms. Without enforcement teeth or the staff to run the program, experts warn the order may serve as little more than a press release—a gesture that reassures no one while changing nothing about how or when powerful AI models hit the market.