Trump Signs Executive Order Establishing Voluntary AI Security Review Framework

WASHINGTON: U.S. President Donald Trump on Tuesday signed an executive order creating a new framework for reviewing the national security implications of advanced artificial intelligence systems, aiming to balance technological innovation with emerging security concerns.

The executive order establishes a voluntary process under which developers of the most advanced AI models may submit their systems for federal review up to 30 days before public release. According to the order, the initiative is designed to help identify and address potential national security risks associated with rapidly advancing AI technologies.

“Advanced AI capabilities make our Nation stronger, but also introduce new national security considerations that require coordinated action across executive departments and agencies,” the order states.

The move comes less than two weeks after Trump postponed a planned White House signing ceremony over concerns that an earlier version of the policy could hinder the United States’ competitive advantage in artificial intelligence. At the time, Trump emphasized the importance of maintaining America’s technological lead over global rivals, particularly China.

The White House clarified that the new framework applies only to developers of cutting-edge AI systems, often referred to as “frontier labs,” and does not constitute broad government oversight of all AI models. In a statement, the administration said the order creates a mechanism for frontier AI companies to voluntarily share advanced cyber-related models to help protect critical infrastructure and strengthen government cybersecurity capabilities.

The administration further stressed that the policy is intended to avoid excessive regulation that could negatively affect innovation and free speech.

Several leading U.S. AI companies, including OpenAI, Google and Anthropic, have been involved in discussions surrounding the initiative. Executives from some of these firms had reportedly been expected to attend the original signing event in May.

Industry reaction to the order has been largely positive. Anthropic described the measure as an important step toward strengthening American leadership in artificial intelligence and expressed its willingness to work with the administration on implementation. OpenAI and Google also welcomed the framework, highlighting the need for responsible safety standards as AI capabilities continue to expand.

Chris Lehane, OpenAI’s chief global affairs officer, said effective AI safety frameworks should be developed through democratic institutions, informed by technical expertise and broad stakeholder participation to promote accountability and public trust.

However, some policy experts have raised concerns about the directive. Juan Londoño of the Cato Institute praised the voluntary nature of the program but questioned how authorities would determine which AI systems qualify for review and which trusted partners would receive early access to sensitive technologies.

The order follows growing debate over the cybersecurity implications of increasingly powerful AI systems. Attention intensified after Anthropic unveiled its advanced AI model, Claude Mythos, earlier this year, prompting discussions among government officials, financial institutions and technology leaders about the potential security risks posed by highly capable AI tools.

Meanwhile, Democratic Senator Mark Warner welcomed the administration’s effort but criticized it for revisiting issues that had previously been addressed under former President Joe Biden. Trump had rescinded several AI-related policies introduced during Biden’s administration shortly after returning to office.

The new executive order signals the administration’s effort to strengthen national security preparedness while preserving the pace of innovation in the rapidly evolving artificial intelligence sector.