President Donald Trump signed an executive order Tuesday to manage oversight of artificial intelligence. This followed the postponing of a previous White House ceremony due to concerns that a similar policy might hinder the U.S.’s technological advancement.
The new order provides a structure for the federal government to evaluate national security risks of advanced AI systems for up to 30 days before public release. AI developers can participate voluntarily. Trump emphasized the importance of AI for national strength but acknowledged new national security concerns needing coordinated government action.
Uncertainties remain regarding differences between the current order and the one Trump chose not to sign on May 21. The 30-day review period is shorter than some industry expectations, aimed at avoiding excessive burdens on a rapidly evolving and competitive field.
Previously, Trump canceled a meeting with tech executives, citing dissatisfaction with an earlier version of the order, saying, “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.”
The directive focuses on voluntary collaboration with U.S.-based tech firms like Anthropic, OpenAI, and Google, known as “frontier labs” due to their work on advanced AI systems. Several companies planned to participate in the signing event, which Trump executed later without ceremony.
The White House clarified the executive order facilitates voluntary sharing of advanced cyber models with frontier labs to secure critical infrastructure and enhance governmental cyber defenses. It stressed oversight will not extend to all new models to prevent overreach.
Juan Londoño of the Cato Institute called the order imperfect but a positive step toward preparing the nation for advanced AI releases. He praised its voluntary nature but noted concerns about vagueness in governmental scrutiny led by the National Security Agency (NSA) and the selection of “trusted partners.” Londoño warned about potential misuse of policy against tech firms like Anthropic.
The cybersecurity directive follows an announcement about Anthropic’s AI model, Claude Mythos, amid legal disputes with the administration. The model’s capabilities prompted urgent discussions with Wall Street leaders concerning security threats. Anthropic has restricted Mythos access to chosen partners but expanded this to 150 additional organizations.
Anthropic and other stakeholders like OpenAI and Google hailed the order as a key move to bolster America’s AI leadership. They stressed the necessity for safety frameworks guided by technical expertise and democratic engagement to foster accountability and trust.
Democratic Senator Mark Warner supported Trump’s policy, noting the administration finally acknowledged the need for structured oversight after dismantling earlier measures.
—AP Technology Writer Barbara Ortutay contributed.