Kartikay Singh
The United States Patent and Trademark Office (“USPTO”) has rescinded its Biden-era guidance concerning inventorship in the context of artificial intelligence (“AI”), reaffirming the long-standing statutory position that only natural persons may qualify as inventors under U.S. patent law. Such a position was also reasserted in Thaler v. Vidal [43 F.4th 1207 (Fed. Cir. 2022)]. As AI systems are not legal persons, they are categorically excluded from inventorship consideration.
On November 28, 2025, the USPTO issued Revised Inventorship Guidance for AI-Assisted Inventions (“Revised USPTO Guidance”), superseding its previous guidance issued on February 13, 2024, titled “Inventorship Guidance for AI-Assisted Inventions” (“Previous USPTO Guidance”). The Previous USPTO Guidance relied heavily on the judicially developed “significant contribution” framework as laid down in Pannu v. Iolab Corp. [155 F.3d 1344, 1351 (Fed. Cir. 1998)] (“Pannu factors”) to assess whether an individual qualified as an inventor. That framework, however, was expressly premised on human-to-human collaboration and proved ill-suited for evaluating AI-assisted inventions.
Under the Previous USPTO Guidance, inventorship required a natural person to make a significant contribution to the conception of each claimed invention, thereby excluding mere prompting, problem formulation, or post-hoc recognition of AI-generated output from qualifying as inventive acts. While this guidance clearly reaffirmed the necessity of human conception, it left unresolved the critical question of how “significant contribution” should be evaluated where AI systems are significant contributors to the inventive process.
On the other hand, the Revised USPTO Guidance represents a material shift in approach. It expressly abandons the “significant contribution” inquiry in the context of AI-assisted inventions and recharacterizes AI as a mere tool that is “analogous to laboratory equipment, computer software, research databases, or other conventional aids” and merely used during the inventive process. In doing so, the USPTO effectively resets the analytical framework for determining inventorship for AI-assisted inventions.
This policy shift is notable for at least two reasons. First, it permits human inventors to attribute inventive features or steps generated with the assistance of AI to themselves, provided that AI is treated solely as a tool rather than a co-inventor. Second, the Revised USPTO Guidance reiterates that AI systems are not legal persons and that a uniform inventorship standard applies across all statutory categories of patentable subject matter, including utility, design, and plant patents. The Revised USPTO Guidance thus neither creates carve-outs nor provides special treatment to AI-assisted inventions.
The Revised USPTO Guidance further clarifies that priority claims must comply with U.S. inventorship requirements. Accordingly, a U.S. patent application may not validly claim priority from a foreign application that names an AI system as an inventor. Priority may be maintained only where at least one common human inventor is identified in both the foreign priority application and the corresponding U.S. application. A foreign application listing an AI system as the sole inventor cannot serve as a valid priority basis before the USPTO.
Critical Assessment
The Revised USPTO Guidance appears to be a deliberate policy effort to adopt a more patent-friendly posture toward AI-assisted inventions. Notably, it expressly discards Pannu factors, which had previously imposed a heightened evidentiary burden on human inventors, as a mechanism for assessing relative human contribution vis-à-vis AI output.
At the same time, the Revised USPTO Guidance implicitly distinguishes between disputes involving multiple human inventors and those involving human-AI interaction. While the Pannu factors may remain relevant in determining inventorship among human collaborators, they no longer govern the analysis between a human inventor and an AI system.
Under the revised framework, so long as a human conceives the definite and permanent idea of the invention and employs AI merely as a tool to implement or refine that idea, the human may properly be named as the sole inventor. This conception is now effectively presumed under a “presumption of inventorship” subject, of course, to compliance with substantive patentability requirements, including subject-matter eligibility under 35 U.S.C. § 101.
One additional aspect of the Revised USPTO Guidance warrants scrutiny. While the USPTO’s analogy of AI systems to databases, laboratory instruments, or conventional software may be legally tenable, it is arguably technically flawed, given the autonomous and generative capabilities of modern AI systems.
Additionally, the Revised USPTO Guidance exhibits several significant lacunae:
- Attribution of Inventive Features
Each named inventor must execute an oath or declaration attesting to having conceived the claimed invention. The Revised USPTO Guidance creates a position that may be contrary in law in which a human inventor may be required to claim inventorship over features substantially generated by AI, potentially conflicting with traditional notions of conception under U.S. patent law.
- Conception versus Reduction to Practice
The Revised USPTO Guidance presumes that inventorship hinges entirely on human conception and that AI-generated outputs are necessarily conceived and reduced to practice by the human user. This assumption oversimplifies the doctrinal distinction between conception and reduction to practice in the context of generative AI.
- Absence of Evidentiary Standards
The Revised USPTO Guidance is silent on any specific standard, procedure, or disclosure obligation for establishing human conception in AI-assisted inventions, leaving applicants and practitioners without meaningful guidance on how to sufficiently establish the same.
- Prompt Engineering as Inventive Activity
The Revised USPTO Guidance implicitly assumes that conception occurs prior to AI output generation, thereby suggesting that prompt engineering may constitute an inventive contribution. This raises an unresolved doctrinal question: whether prompt engineering should be treated as inventive activity or merely as routine instruction of a computational tool.
- Tool versus Creator Analogy
Finally, the Revised USPTO Guidance does not meaningfully address whether AI is more appropriately analogised to a creator (e.g., a painter) or an instrument (e.g., a paintbrush), nor does it identify any threshold at which an AI system might cease to function as a mere “tool” for purposes of inventorship determination.