In the wake of the Supreme Court's 2024 decision dismantling forty years of Chevron deference, American law finds itself at a crossroads over who should govern artificial intelligence — and how. The old arrangement allowed federal agencies to quietly foreclose state-level accountability without having to justify why their rules were sufficient; the new one risks replacing bureaucratic overreach with judicial improvisation. What the moment calls for is neither deference nor dominance, but a discipline of reasoning: agencies earning authority through the quality of their arguments, and states re