Over the last several months, the United States Patent and Trademark Office (“USPTO”) publicized its attention to the impact of artificial intelligence on U.S. patent practice. On February 6, 2024, the USPTO issued its guidance document (“Board Memo,” available here) related to the potential use of AI in legal proceedings before the Patent Trial and Appeal Board and the Trademark Trial and Appeal Board (collectively, “the Board”). Shortly thereafter, on February 12, 2024, the USPTO issued its Inventorship Guidance for AI-Assisted Inventions, 89 Fed. Reg. 10043 (Feb. 13, 2024) (“Guidance,” available here), which clarified how the USPTO will treat inventorship of inventions made using AI and further expanded on the Board Memo’s discussion of practitioners’ duties when practicing before the USPTO with the assistance of AI.

The Guidance applies to all patents issuing from applications filed before, on, or after February 13, 2024, and clarifies that AI cannot be a sole or joint inventor of patents. The USPTO is seeking comments on the Guidance, which must be received in writing by May 13, 2024. The Guidance’s discussion accords with the USPTO’s decisions denying petitions seeking to name the Device for the Autonomous Bootstrapping of Unified Sentience (“DABUS”; an AI system) as an inventor of two patent applications related to two devices that were created solely by DABUS. The USPTO’s denials were later affirmed by the Court of Appeals for the Federal Circuit. Non-U.S. jurisdictions have treated the issue similarly and have devised similar tests as set forth in the Guidance. Further description of these topics follows below.

The Guidance’s Treatment of AI-Assisted Inventions

The Guidance clarified that inventions made using AI would not be categorically unpatentable, but AI itself cannot properly be named as a sole inventor or a joint inventor.

Turning first to the issue of inventorship, the Guidance summarized the USPTO’s decisions related to AI inventorship, which was upheld by the United States District Court for the Eastern District of Virginia and eventually affirmed by the Federal Circuit in Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022). There, the court looked to 35 U.S.C. § 100(f), which defines an inventor as “the individual” or “individuals” who invented or discovered the invention. Following Supreme Court precedent reasoning that an individual is ordinarily a human being, the court held “that only a natural person can be an inventor, so AI cannot be.” Thaler, 43 F.4th at 1213. Consequently, the Thaler decision closed the door on the possibility of AI as an inventor.

However, while AI may not be an inventor, the use of AI by a natural person does not make the invention categorically unpatentable for improper inventorship. The Guidance clarified that properly identifying the inventors of a patent made using AI requires identifying the natural persons that significantly contributed to the claimed invention.

Determining whether a natural person should be named as an inventor as a result of a significant contribution to the claimed invention requires analyzing the three “Pannu factors” established in Pannu v. Iolab Corp., 155 F.3d 1344 (Fed. Cir. 1998). In light of the Pannu factors, the Guidance acknowledged that there is no bright-line rule to ascertain “significant contribution,” but provided a non-exhaustive list of factors to help inform the decision (Guidance at 10048-49):

1. “A natural person’s use of an AI system in creating an AI-assisted invention does not negate the person’s contribution as an inventor” so long as the person “contributes significantly to the AI-assisted invention.”
2. “Merely recognizing a problem or having a general goal or research plan to pursue does not rise to the level of conception . . . . However, a significant contribution could be shown by the way the person constructs the prompt in view of a specific problem to elicit a particular solution from the AI system.”
3. “Reducing an invention to practice alone is not a significant contribution that rises to the level of inventorship.” But a person taking the output of an AI system and [making] a significant contribution to the output to create an invention” may be an inventor. Alternatively, a person that “conducts a successful experiment using the AI system’s output could demonstrate that the person provided a significant contribution to the invention even if that person is unable to establish conception until the invention has been reduced to practice.”
4. “A natural person who develops an essential building block from which the claimed invention is derived may be considered to have provided a significant contribution to the conception of the claimed invention even though the person was not present for or a participant in each activity that led to the conception of the claimed invention.”
5. The USPTO provided the caveat that: “Maintaining ‘intellectual domination’ over an AI system does not, on its own, make a person an inventor of any inventions created through the use of the AI system.”

The USPTO’s Guidance Accords With International Jurisdictions

To date, the United Kingdom’s Intellectual Property Office (“UKIPO”), the USPTO, the German Patent and Trade Mark Office (“DPMA”), IP Australia, the Intellectual Property Office of New Zealand (“IPONZ”), and the Korean Intellectual Property Office (“KIPO”) have all concluded that an inventor must be a natural person after reviewing the DABUS patent applications in their respective jurisdictions. The Japanese Patent Office and the China National Intellectual Property Administration have also indicated that an inventor must be a natural person. To date, only South Africa has allowed the DABUS patent applications to issue as patents, but this may simply be because South Africa has a registration system that does not formally examine applications before they are registered as patents. Thus, it remains to be seen whether South Africa’s Intellectual Property Office would uphold registration if it were opposed. Nonetheless, there appears to be a growing consensus worldwide that inventors must be human beings.

The European Patent Office (“EPO”) has also discussed AI-assisted inventions. The EPO’s discussion of AI-assisted inventions identified three types of AI-inventions, determining that the first two categories are capable of having a human inventor:

1. human-made inventions using AI for the verification of the outcome;
2. inventions in which a human identifies a problem and uses AI to find a solution; and
3. AI-made inventions, in which AI identifies a problem and proposes a solution without human intervention.

The EPO has explained that the first two categories can have a human inventor because they represent instances where a human is creating the invention using their own creative activity with AI as a tool, whereas the third cannot have a human inventor because AI is inventing independent of human intervention.

The USPTO’s Guidance Explains That Practitioners’ Duties Are Largely Unchanged but Cautions Practitioners When Employing AI in Their Practice

The Board Memo noted recent concerns around practitioners’ use of AI, such as the phenomenon referred to as “hallucination,” where AI presents inaccurate or nonsensical information as fact. In one instance, hallucination led to citation to non-existent cases by an attorney in a filing submitted to a district court. Nonetheless, the USPTO did not establish any new rules or duties related to proceedings in front of the Board, concluding that “the agency’s existing rules are adequate to address the challenges that the USPTO is likely to face.” (Board Memo at 5).

The Guidance also made no changes to practitioners’ duties of disclosure or reasonable inquiry. (Guidance at 10049-50). However, the Guidance offered two cautionary notes that: (1) “special care should be taken by those individuals subject to [the duty of disclosure] to ensure all material information is submitted to the USPTO to avoid any potential negative consequences; and (2) “patent practitioners who are preparing and prosecuting an application should inquire about the proper inventorship,” which in the context of AI-assisted inventions, “could include questions about whether and how AI is being used in the invention creation process.” The Guidance otherwise noted that it was making no changes to the process to correct inventorship or to the oath or declaration practice for naming inventors.

Conclusions and Takeaways

Going forward, patent practitioners will have to be diligent when working with inventors who are using AI. Given the ubiquity of AI in coming years, practitioners should inquire about the process the inventors used to create an invention, along with whether and how the inventors used AI. Practitioners should determine whether and to what extent an AI system was used for conception and reduction of the invention to practice, whether and to what extent a natural person relied on the outputs from the AI system, and how these facts dictate the overall inventorship determination.

Furthermore, the Guidance noted that pursuant to President Biden’s Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence (Oct. 30, 2023), the USPTO Director shall issue additional guidance within 270 days to address other considerations at the intersection of AI and IP, which could address issues related to novelty, written description, and patent eligibility. In any event, there will be significant legal development at the intersection of AI and IP, particularly related to preparing, filing, transacting, and litigating AI-related patents.