US President Donald Trump signed an executive order Tuesday to establish government oversight of emerging artificial intelligence technologies.
Section 3 of the order calls for developing and maintaining a “classified benchmarking process to assess the advanced cyber capabilities of AI models.” Specifically, the administration wants to determine which models constitute a “covered frontier model,” a designation for the most powerful, high-energy models available. Departments included in Section 3 are War/Defense, Treasury, Homeland Security and Commerce.
Furthermore, the order creates a voluntary framework whereby AI developers can “provide the Federal Government with access to covered frontier models.” This is to take place 30 days before they plan to release the models to the public.
The Trump administration has previously criticized AI regulations, and, in fact, continues to do so. The White House is promoting this order as balancing innovation and national security. Officials distinguished the order from the Biden administration’s policy, claiming the previous administration sought out a top-down approach. Trump, on the other hand, works “hand-in-hand with American industry,” per the White House website.
Representative Don Beyer (D-VA), co-chair of the bipartisan Congressional Artificial Intelligence Caucus, was critical of the order. He said the order “continues to allow companies to release powerful models to the public without undergoing meaningful evaluation or review.” Additionally he criticized the “absence of clear guidance on how identified threats would be addressed.”
The order can be seen as step towards defending against AI agents and other cyberattacks. Its defensive focus seeks to change the scenario Vectra AI CEO Hitesh Sheth warned against, in which attackers have an advantage. To avert this, he recommended AI expertise at the top of both enterprises and government.
Most elements of the order will take effect in 30 days.
