A universal requirement of patent law is that inventors must disclose information about how to implement their inventions. Increasingly, however, portions of patents are written not by inventors (or their lawyers), but by generative artificial intelligence (AI). The burgeoning market for AI-assisted patent drafting may lower patenting costs, but it will also exacerbate existing disclosure problems. AI will amplify errors in both improperly granting poorly disclosed patents and improperly denying patents on the basis of earlier AI-generated references that do not actually disclose those inventions. In recognition of these issues, the US Patent and Trademark Office (USPTO) recently issued guidance requiring human review of AI-drafted submissions. But this will not be enough. Increasing enforcement of existing disclosure rules and heightening requirements to limit granting patents to those who have performed real inventive work would both equip patent offices for this challenge and improve the patent system overall.