Navigating Inventorship of AI-Assisted Inventions: USPTO’s Guidance and Implications

This article discusses the February 13, 2024 guidance[1] issued by the United States Patent and Trademark Office (USPTO) regarding the inventorship of artificial intelligence (AI)-assisted inventions. While this guidance marks a significant advancement in AI policy, it is not the first initiative by the USPTO in this area. The USPTO has been addressing concerns related to AI-assisted inventions for several years, specifically requesting comments and feedback on issues involving inventorship and AI-assisted inventions, which have influenced the content of the February 13, 2024 guidance. Three main points of consideration related to the inventorship of AI-assisted inventions are discussed below.

First, the USPTO clarified that while AI-assisted inventions are not automatically unpatentable, an AI system itself cannot be designated as an inventor or co-inventor. Affirming the Federal Circuit’s holding in Thaler,[2] the USPTO reiterated that only a natural person may be an inventor, but confirmed that using an AI system in creating an invention does not diminish the natural person’s role as an inventor.

Second, the guidance indicates that a natural person who utilizes an AI system may qualify as an inventor if the natural person makes a “significant contribution” to the invention and has “recognition and appreciation” of the invention. Merely recognizing and presenting a problem to an AI system, or appreciating the output derived from an AI system, will likely not rise to the level of inventorship. That being said, a person may be able to demonstrate a significant contribution by skillfully formulating the prompt for the AI system to elicit a particular solution from the AI system, or significantly experimenting on or enhancing the AI system’s output. Similarly, simply reducing the AI system’s output to practice, owning, training, or maintaining the AI system used in the invention will likely not meet the threshold of inventorship. But, a natural person who develops an “essential block” of the invention could be considered an inventor by significantly contributing to the invention. For instance, a person may demonstrate a significant contribution by designing or training an AI system in a specific way to elicit a particular solution.

The Pannu[3] factors are the standard for determining whether a natural person has made a “significant contribution” to the invention. Specifically, each named inventor must satisfy the three Pannu factors. Under the Pannu factors, a person makes a significant contribution to an invention if they: (1) contributed in some significant manner to the conception of the invention; (2) made 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) did more than merely explain the well-known concepts and/or the current state of the art. And because inventorship is determined on a claim-by-claim basis, an inventor must satisfy the Pannu factors for each claim. Importantly, the failure of a person to meet any one of these Pannu factors prevents that person from being designated as an inventor.

Educating inventors about these criteria can help them understand how their inventions are evaluated under patent law, making it easier for their patent filings to meet USPTO AI requirements. For example, inventors may consider drafting prompts for the AI system carefully, as well as performing additional experiments and modifying the AI generated output to satisfy the Pannu factors. In addition, applicants may want to consider, early in the drafting process, whether each inventor satisfies the Pannu factors for each element of each claim. Carefully aligning inventor contributions with the claims can strengthen the patent application and defend against challenges during examination or litigation. For example, if an applicant is unsure if a particular claim element meets the Pannu factors, the applicant may want to place that particular claim element in a dependent claim, rather than an independent claim, to protect the independent claims against potential validity challenges. Patent owners may also want to consider carefully reviewing patent portfolios that include inventions created with AI assistance. They may want to examine each claim to assess whether any action, such as filing a reissue application, is necessary to amend the claims to mitigate potential challenges to the patent’s validity.

Third, the guidance confirms that the existing duties of disclosure and reasonable inquiry still apply to AI-assisted inventions. As part of the duty of disclosure, an applicant has a duty to disclose all information that is material to patentability, which includes information that may demonstrate that a named inventor did not make a significant contribution to the invention.  Importantly, if the facts or evidence indicate that the named inventor(s) did not significantly contribute to the claimed invention, a rejection under 35 U.S.C. § 101 and 115 may be warranted. The duty of reasonable inquiry requires the applicant to asses if the contributions of a natural person rise to the level of inventorship. Inventors may consider taking detailed notes of human involvement during the invention process, for example, the specific role of each natural person in the inventive process, to help establish inventorship and protect against challenges. Inventors may also consider noting the specific prompts and information presented to the AI system, as well as any modifications or changes made to the output. Applicants should be cognizant of these requirements when determining inventorship and drafting applications.

As the intersection of AI and patent law continues to evolve, these principles published by the USPTO will play a crucial role in shaping future innovation and legal precedent.

[1] 89 Fed. Reg. 10043 (Feb. 13, 2024). This guidance has not been tested by the courts and is intended to offer informative assistance to the public, such as clarifying responsibilities under existing laws or regulations. This guidance does not impose obligations on the public directly.

[2] Thaler v. Vidal, 43 F.4th 1207, 1213 (Fed. Cir. 2022).

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