Rogue AI Patents and the USPTO’s Rejection of Alice Joshua A. Z

AI inventions, from ChatGPT to self-driving taxis, have taken the world by storm. Many of these inventions are protected by patents, and the number of AI patents is rapidly growing. Yet a large number of AI patents are flawed, prone to invalidation in court. This Comment argues that many AI patents were granted on legally flawed grounds, pursuant to United States Patent and Trademark Office (USPTO) regulations that systematically departed from controlling case law. The existence of these flawed patents poses a growing problem. Courts may invalidate the patents, upsetting expectations of an important, nascent industry. On the other hand, courts may acquiesce to the USPTO’s leniency, which could have the perverse effect of further unsettling the law, increasing examination uncertainty, and proliferating bad patents. This Comment asks, in light of the policy of the patent system, which AI inventions ought to receive patents. It concludes that AI methods and models should be patent eligible because they are likely to be incentivized by patents and unlikely to chill follow-on innovation. This Comment further argues that both the USPTO’s guidance and much of the Federal Circuit’s recent eligibility case law are inconsistent with finding these inventions patent eligible. However, the Federal Circuit demonstrated an understanding of eligibility that would allow patents for many AI methods and models in its 2016 McRO, Inc. v. Bandai Namco Games America Inc. decision. This Comment concludes by advocating that the Federal Circuit explicitly apply the holding of Bandai to hold that an AI invention is patent eligible at the first opportunity in order to settle the law while granting the benefits of patents only to deserving inventions.