In a significant move, U.S. President Donald Trump has signed an executive order aimed at increasing government oversight of advanced artificial intelligence technologies. The directive primarily seeks to implement a system of classified benchmarking to assess the cyber capabilities of what are termed as “covered frontier models.” These are the most sophisticated AI models currently available, known for their high energy capacity. Notably, key governmental departments such as War/Defense, Treasury, Homeland Security, and Commerce are involved in this initiative.
This executive order also establishes a voluntary framework for AI developers, allowing them to provide access to these frontier models to the federal government 30 days prior to their public release. The emphasis on voluntary cooperation underlines the administration’s effort to balance innovation with national security. In contrast to previous administrations, the Trump White House is promoting this approach as a collaborative effort with American industry, distancing itself from the top-down methods of earlier policies.
Critics, however, have voiced concerns. Congressman Don Beyer, co-Chairman of the Congressional Artificial Intelligence Caucus, argues that the order lacks stringent evaluation or review processes before companies release powerful models to the public. He also highlights an absence of clear guidelines for addressing identified threats, raising concerns about the potential risks of unregulated AI deployment.
While the Trump administration’s stance has been notably skeptical of strict AI regulations, this executive order marks a step toward enhancing defenses against AI-driven cyberattacks. The focus on cybersecurity aims to shift the balance of power, which currently favors attackers, as observed by Vectra AI CEO Hitesh Sheth. The inclusion of AI expertise at the enterprise level, and within government departments, is seen as crucial to mitigating these threats.
This order, which will become effective within 30 days, signals a strategic shift in the U.S. approach to handling the rapid evolution of AI technologies. The administration’s position and the reactions it provokes are outlined in a further report available here.