Trump Plans to Sign Executive Order Granting Oversight of A.I. Models

President Trump will sign an executive order on Thursday establishing government oversight of artificial intelligence models before their public release, marking a departure from the White House’s previous hands-off stance. The order empowers the Office of the National Cyber Director and other agencies to develop a two-month process for evaluating AI models, focusing on detecting security risks that could threaten banks, utilities, and critical infrastructure. The proposal requires major AI companies to voluntarily share their models 14 to 90 days before release, allowing the government to assess potential vulnerabilities. Officials also plan to create a centralized vault for reporting and patching security flaws discovered through AI testing. The shift follows concerns that unchecked AI advancements could pose national security risks, particularly after Anthropic’s Mythos model demonstrated its ability to uncover software vulnerabilities. The decision stems from fears that adversarial actors—including China—could exploit AI-driven cybersecurity weaknesses, turning them into tools for espionage or sabotage. Until recently, the Trump administration had prioritized AI innovation over regulation, viewing it as a key advantage in the U.S.-China technological competition. However, the release of Mythos last month intensified debates within the government, with officials arguing that proactive oversight was necessary to prevent catastrophic breaches. Negotiations for the executive order began in early May, with White House officials consulting AI executives ahead of a planned summit between Trump and Chinese President Xi Jinping on May 14. The U.S. and China are engaged in an AI arms race, competing over advancements in language models, semiconductor technology, and AI-powered weapons systems. The order reflects internal tensions between officials advocating for regulation to safeguard national security and those resisting interference in corporate-led innovation. The White House did not immediately respond to requests for comment, nor did Anthropic, OpenAI, Google, or Microsoft. The move signals a broader effort to balance AI’s rapid growth with the need to mitigate emerging threats, particularly as foreign adversaries accelerate their own AI capabilities.