After the Thaler v. Vidal ruling established that an AI cannot be a named inventor, a practical question remained: can a human still patent an invention that was developed with substantial help from an AI system? On February 13, 2024, the US Patent and Trademark Office answered with formal Inventorship Guidance for AI-Assisted Inventions, published in the Federal Register (89 FR 10043). The guidance, accompanied by worked examples, told examiners and applicants how to handle inventions where AI played a role.
The core principle was that AI-assisted inventions are not categorically unpatentable, so long as one or more natural persons made a “significant contribution” to the claimed invention. The USPTO drew on the existing Pannu factors - the long-standing test for joint inventorship - to assess whether a human’s involvement was meaningful enough. Crucially, the guidance warned that merely owning, overseeing, or maintaining “intellectual domination” over an AI system is not enough; recognizing a problem, building a workable prompt that elicits a specific solution, or making a meaningful technical contribution to the result can be. The framework was deliberately centered on rewarding human ingenuity, consistent with the constitutional purpose of patents.
The guidance was a snapshot of evolving policy rather than the last word; the USPTO took public comments and later revised its approach in 2025.
Why business readers should care: companies using AI in their R&D pipelines need to document the specific human contributions to each invention. Under this guidance, sloppy records of who actually contributed - versus what the AI generated - can put a patent’s validity at risk.