ARTICLE
11 December 2025

USPTO Issues New Guidance On AI-Assisted Inventorship, Rescinding Its 2024 Guidance

OB
Osha Bergman Watanabe & Burton LLP

Contributor

OBWB is a full-service intellectual property law firm with a global reach. Our expertise in intellectual property law and IP-intensive industries coupled with our client-focused culture enables us to deliver a premium level of quality, value, and service. At our strategically located offices in the United States, France, Japan, China, and the UAE, we offer clients assistance with patents, copyrights, trademarks, trade secrets, and unfair competition. At OBWB we strategize, prepare and file, negotiate, license, enforce and defend, and facilitate monetizing intellectual property assets.
On November 28, 2025, USPTO Director John Squires issued new guidance on AI-assisted inventorship and explicitly rescinded the February 13, 2024, guidance that had been issued by his predecessor Director Vidal.
United States Intellectual Property

On November 28, 2025, USPTO Director John Squires issued new guidance on AI-assisted inventorship1 and explicitly rescinded the February 13, 2024, guidance that had been issued by his predecessor Director Vidal.2 At issue is what human activity rises to the level of inventorship when a human uses an AI tool in a manner that results in an invention.

It is well understood that AI systems cannot be named as inventors; this is not in question in either version of the guidance. The issue is that, as AI systems continue to become more powerful, there may be situations where a potentially patentable invention results from a human using such a system as a tool of innovation and where that human did little more than enter a prompt. Indeed, an entire spectrum of relative human vs AI contribution can be envisioned, from minimal contribution by the human to minimal contribution by the AI system. Further, it matters not only how the AI system was used but also how the results were used. Did the human recognize something unexpected in the results? Was a result selected and tested? Altered or refined? These and many other possible scenarios make this a complicated area of the law where, at least in the United States, there is currently little clarity.

The 2024 Guidance attempted to shed some light on the topic by drawing an analogy to the existing law of joint inventorship. Specifically, under U.S. law, the determination of whether a person is an inventor in a joint inventorship situation (with another human) rests on whether that person made a "significant contribution" to the claimed invention. "Significant contribution" is, in turn, defined by the "Pannu factors" established by an eponymous 1998 Federal Circuit decision.3 The Pannu factors require, for a person to be a coinventor, that the person:

(1) contribute in some significant manner to the conception or reduction to practice of the invention;

(2) make a contribution to the claimed invention that is not insignificant in quality, when that contribution is measured against the dimension of the full invention; and

(3) do more than merely explain to the real inventors well-known concepts and/or the current state of the art.

This approach taken in the 2024 Guidance had some appeal because it gave a framework, based on existing law, upon which to evaluate difficult inventorship scenarios that may arise in the AI use context. The 2024 Guidance was also welcomed for confirming that an invention is patentable so long as there is at least one true human inventor; meaning that contribution by an AI system, no matter how significant, would not negate patentability. However, commentators (including the author of this article) were quick to note that application of the Pannu factors to the human/AI situation is not entirely apt. The Pannu factors, which are used to evaluate the significance of a contribution by one person to a joint invention, inherently require there be at least one additional human who has also made such contribution. No such additional human exists in the "human using AI" scenario.

The recently-issued 2025 Guidance rescinds the 2024 Guidance in its entirety, making quick work of the Pannu factors approach:4

The Pannu factors only apply when determining whether multiple natural persons qualify as joint inventors. Pannu is inapplicable when only one natural person is involved in developing an invention with AI assistance because AI systems are not persons and therefore cannot be "joint inventors" so there is no joint inventorship question to analyze.5

The 2025 Guidance clarifies that the same legal standard of inventorship applies to all inventions and there is no "separate or modified standard for AI-assisted inventions."6 Referencing the conventional U.S. test for inventorship, the 2025 Guidance explains,7

Determining inventorship is highly fact intensive. The question is whether the natural person possessed knowledge of all the limitations of the claimed invention such that it is so ''clearly defined in the inventor's mind that only ordinary skill would be necessary to reduce the invention to practice, without extensive research or experimentation.'' Analysis of conception turns on the ability of an inventor to describe an invention with particularity. Absent such a description, an inventor cannot objectively prove possession of a complete mental picture of the invention at a later time.

While it is helpful to confirm that the same legal standard of inventorship applies to all inventions, the 2025 Guidance does little to give practical guidance for the complex scenarios that the 2024 Guidance attempted, rightly or wrongly, to address. For that, the innovation community must wait for some of these scenarios to be addressed in future court decisions. In the meantime, we highly recommend that inventors who use AI tools as part of their inventive process keep detailed records of their contributions, including how systems were trained, what prompts were used, and what post-result analysis or modification may have been done. It seems likely that future cases in this area will turn, at least in part, on what the human inventors can objectively prove they had as their "mental picture" at the time the invention was made.

Footnotes

1 https://www.federalregister.gov/documents/2025/11/28/2025-21457/revised-inventorship-guidance-for-ai-assisted-inventions

2 https://www.federalregister.gov/documents/2024/02/13/2024-02623/inventorship-guidance-for-ai-assisted-inventions

3 Pannu v. Iolab Corp., 155 F.3d 1344, 1351 (Fed. Cir. 1998)

4 90 FR 54636 (Nov. 28, 2025) (citations omitted).

5 There is a subtle conflation of concepts in this statement. U.S. law is clear that only natural persons can be "inventors" as that term is used in the statute and hence only natural persons can be listed as inventors on a patent application. Thaler v. Vidal, 43 F.4th 1207, 1212 (Fed. Cir. 2022). What the Thaler case did not answer -- because it was not at issue in the case -- is whether AI systems can make a contribution to an invention that, if that same contribution were made by a human, would amount to inventorship by that human. This is an open question about which reasonable minds will disagree.

6 Id.

7 Id. at 54637 (citations omitted).

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.

[View Source]

Mondaq uses cookies on this website. By using our website you agree to our use of cookies as set out in our Privacy Policy.

Learn More