Pentagon threatens export controls for non-compliant AI labs

Pentagon Undersecretary Emile Michael stated that the government will use the full range of executive powers, including export controls, if AI labs do not comply with the new AI task force. Michael, one of four senior leaders on the task force which met for the first time last week, expressed concern about Chinese open-source models becoming dominant in the U.S., noting that these companies are subject to the rule of the Chinese government. He suggested the government could audit the weights of open models but preferred a competitive approach to boost domestic developers like Reflection AI and NVIDIA. Regarding the recent case of a Saudi OpenAI employee charged with installing a listening device, Michael urged AI labs to tighten security and promoted an FBI vetting service for hiring foreign nationals. He dismissed the idea that labs need government enforcement to slow down, likening them to drunk drivers begging to be pulled over. Michael stated he does not hold a grudge against Anthropic despite frequent conflicts, asserting his first duty is to the department and American people from a national security standpoint.