Abstract
It has been suggested that a national frontier AI governance strategy should include a comprehensive regime for tracking and licensing the creation and dissemination of frontier models and critical hardware components (“AI Oversight”). A robust Oversight regime would almost certainly require new legislation. In the absence of new legislation, however, it might be possible to accomplish some of the goals of an AI Oversight regime using existing legal authorities. This memorandum discusses a number of existing authorities in order of their likely utility for AI Oversight. The existing authorities that appear to be particularly promising include the Defense Production Act, the Export Administration Regulations, the International Emergency Economic Powers Act, the use of federal funding conditions, and Federal Trade Commission consumer protection authorities. Somewhat less promising authorities discussed in the memo include § 606(c) of the Communications Act of 1934, Committee on Foreign Investment in the United States review, the Atomic Energy Act, copyright and antitrust laws, the Biological Weapons Anti-Terrorism Act, the Chemical Weapons Convention Implementation Act, and the Federal Select Agent Program.
Matthijs Maas
Senior Research Fellow
I am a Senior Research Fellow at the Institute for Law & AI, working on adaptive global governance approaches for AI.