Generative AI and Patent Law: Challenges to Disclosure Requirements in Modern Innovation Systems

The advent of generative artificial intelligence as an innovative technology brings into question some of the very foundations upon which patent disclosure obligations rest. Patenting involves making sufficient disclosure about the innovation to the public in exchange for the right to secure a patent on the innovation – an exchange based on the idea that the innovation results from the workings of human minds, which can describe the innovation in a manner intelligible to others having ordinary skill in the pertinent area of technology. The use of generative AI technologies threatens all of these fundamental premises, because such systems operate under black box conditions, have stochastic outputs, dispute any claim of contribution to inventiveness, and embody information that is simply unknowable to human observers. In this article, I offer an analysis of the challenges posed to patent disclosure requirements – including written description, enablement, best mode, and inventorship disclosure – by generative AI technologies in light of current case law, patent office guidance, and recent empirical data concerning patent filings. The essay further asserts that an adjustment to doctrines alone cannot solve the inherent epistemological conflict between generative AI and existing patent disclosure law, and that there is dire need for an international approach to dealing with innovations produced by AI technology.