President Donald Trump privately signed a scaled-back executive order Tuesday giving AI and tech companies a brief reprieve to allow the federal government to independently review their AI models.
The order comes after the federal government repeatedly consulted with major AI companies, including Anthropic, about cybersecurity. In April, the company announced it would limit the release of Claude Mythos over concerns that the AI model was too adept at finding software and cybersecurity vulnerabilities. Anthropic, which secretly filed for an initial public offering, announced last week that it would soon release a Mythos-level model to customers.
Under a scaled-back executive order, leading AI model makers and tech companies will be able to voluntarily provide the federal government with access to AI models with advanced cyber capabilities up to 30 days before release. A previous executive order published by Politico called for a review period of up to 90 days.
The new executive order emphasizes that any review is not mandatory.
President Trump previously told reporters he delayed signing the order over concerns it could jeopardize U.S. companies’ lead over China in the AI race.
President Trump told reporters on May 21, “We’re leading China. We’re leading everyone,” and “I don’t want to do anything to get in the way of that lead.”