The utilization of artificial intelligence (“AI”) is becoming ubiquitous across a wide range of industry sectors. The biotech industry is no exception.
AI-driven platforms have become increasingly useful to biotech innovators in many areas including in discovering and developing drugs. For example, a 2022 article by McKinsey & Company estimated that nearly 270 companies have utilized AI in drug development with over half of those companies based in the United States. Such companies are using AI to accelerate inherently burdensome tasks in drug development including target identification, target validation, hit identification, lead generation, lead optimization and in other areas. Id.
Utilizing new technologies as tools to develop drugs is by no means unique over the course of history. However, AI is a special case. Never before has there been a technology (such as generative AI) that has had the ability of performing inventive acts similar to the human mind. This capability creates a conflict with longstanding U.S. inventorship law.
Certainly, the legislators that passed the 1952 Patent Act (“Patent Act”), which established the foundation for U.S. patent law for the past 72 years, did not envision the power of AI and its potential to contribute inventing acts to the development of inventions. While there are thousands of cases that have interpreted different aspects of the Patent Act over the decades, courts have yet to establish standards to determine whether AI-developed invention are eligible for patent protection.
This uncertainty in the law presents significant potential exposure for biotech companies that utilize AI to develop drugs, and such companies should take steps to mitigate that exposure. This post provides (1) an overview of the relevant legal landscape for inventorship under U.S. law in the context of AI-developed drugs; and (2) key considerations for drug developers to mitigate their exposure in view of this uncertain legal landscape.
Legal Landscape for Inventorship When Utilizing AI
The Federal Circuit confirmed in Thaler v. Vidal, 43 F.4th 1207, 1210 (Fed. Cir. 2022) that under the Patent Act only a “natural person” (i.e., a human being) can be an inventor on a patent application and AI cannot be an inventor.
The Thaler court explicitly did not address “the question of whether inventions made by human beings with the assistance of AI are eligible for patent protection.” Id. at 1213 (emphasis original). There is therefore an open question in the law regarding what constitutes inventorship when human beings utilize AI to develop new drugs (and other technologies).
In view of this, and pursuant to the October 30, 2023 Executive Order on the Safe, Secure, and Trustworthy Development of Artificial Intelligence, the USPTO issued the Inventorship Guidance for AI-Assisted Inventions (“Guidance”) on February 13, 2024. We posted about this earlier this year.
The stated purpose of the Guidance is to “provide clarity for USPTO stakeholders and personnel … on how the USPTO will analyze inventorship issues as AI systems, including generative AI, play a greater role in the innovation process.” Id. at 10044.
The Guidance recites the standards for inventorship under longstanding U.S. law, i.e., that the inventor is the person who “conceived” of the invention, and that “conception” is “the mental act” of the invention, i.e., “the formation in the mind of the inventor, of a definite and permanent idea of the complete and operative invention …” Id. The Guidance also confirms that under U.S. law, if more than one person contributed to the conception of the invention, each person would be considered a “joint inventor.” Id.
While recognizing that under Thaler an AI system cannot be an inventor, the Guidance also acknowledges that AI systems are able to perform acts that would constitute inventorship if those acts were otherwise performed by a human. Id. at 10046. The Guidance suggest that, in a situation where a human being utilized AI to develop an invention, proper inventorship can only be found when the “natural person” “significantly contributed to the claimed invention….” Id. This, according to the Guidance, is grounded in the (now well-established) factors set forth in Pannu v. Iolab Corp., 155 F.3d 1344, 1351 (Fed. Cir. 1998) used to evaluate joint inventorship.
The Guidance suggests applying the following Pannu factors in the context of AI-assisted inventions in order to determine whether the contribution was “significant,” i.e. did the natural person:
- Contribute in some significant manner to the conception of the invention;
- Make a contribution to the claimed invention that is not insignificant in quality, when that contribution is measured against the dimensions of the full invention; and
- Do more than merely explain to the real inventors well-known concepts and/or the current state of the art.
Id. at 10047 (quoting Pannu v. Iolab Corp., 155 F.3d 1344, 1351 (Fed. Cir. 1998)). The Guidance states that failure to meet any one of these criteria can preclude a person from being named as an inventor. Id.
The Guidance also states that a “natural person must have significantly contributed to each claim in a patent application or patent” and instructs that applicants must “take extra care” to ensure that a significant human contribution was made to each claim on a claim-by-claim basis. Id. at 10048.
The Guidance also states that there is no “bright-line test” regarding whether a natural person’s contribution to an AI-assisted invention was significant and, in addition to the Pannu factors, provides “Guiding Principles” to assist applicants and the PTO as to proper inventorship (see Exhibit A below). Id.
The Guidance also indicates that a patent applicant’s Duty of Disclosure to the USPTO applies to AI-assisted inventions including, inter alia, the disclosure of “evidence that demonstrates a named inventor did not significantly contribute to the invention because the person’s purported contribution(s) was made by an AI system.” Id. at 10049.
Key Inventorship Considerations for AI in Drug Development
Exemplary Applications of AI in Drug Development
As indicated above, AI has been particularly useful to biotech innovators to dramatically reduce time and costs associated with key activities in the drug discovery and development process. By way of example only, as shown below, innovators have used AI to identify targets (e.g., a protein that a future drug will interact with to treat disease); identify lead compounds (e.g., compounds that are predicted to effectively interact with the target after screening large numbers of potential compounds); design novel compounds (e.g., small molecules and proteins) to serve as leads;
optimize lead compounds (e.g., continued optimization of chemical structure, study of biological interactions, and further testing); repurpose drugs (e.g., screening of large numbers of known drugs and predicting utility for diseases that the drug was not previously for); and design clinical trials (e.g., recruitment of suitable patients and monitoring to mitigate dropout rates).
For purposes of this post, we will focus on inventorship issues relating to patents with claims covering new compounds developed through the utilization of AI.
The Inventorship Dilemma for AI-Developed Compounds
Assume the scenario where a scientist is using AI to develop compounds to treat a particular disease (“Disease X”). During that process, the scientist feeds the AI system data relating to all known genus and species compounds known to treat diseases similar to Disease X and prompts the AI system to provide a list of compounds that may be effective in treating Disease X.
After a short period of time, the AI system identifies 10 compounds, 5 known compounds and 5 new compounds (created by the AI system), as potentially effective in interacting with a certain target protein that can regulate Disease X.
The scientist then tests those compounds and determines that 3 of the new compounds show the most promise and pursues further research designating those compounds as lead compounds. New and exciting data is generated that validate the potential efficacy of the compounds. The scientist’s company then files a patent application claiming the new compounds.
Is the scientist a proper inventor of the new compounds (i.e., the compositions of matter)?
Under the current standards, there is no clear answer. On the one hand, it could be argued that the scientist did not contribute to the conception of the novel compounds in a “significant manner” or quality (i.e., the scientist did not conceive of any novel aspect of the claimed structure) and did no more than feed well-known compounds to the AI system, failing to satisfy any of the Pannu factors.
On the other hand, it could also be argued that under the Guiding Principles, the manner in which the scientist “constructed the prompt” of the AI system, made a “significant contribution to the output” of the AI system, or conducted “successful experimentation” and reduction to practice using that output may have been enough to establish that the scientist provided a significant contribution to the conception of the invention.
Moreover, in the context of chemical compounds, conception requires both (1) conception of the compound’s structure and (2) “possession of the operative method of making it.” Amgen, Inc. v. Chugai Pharm. Co., 927 F.2d 1200, 1206 (Fed. Cir. 1991). This is known as the doctrine of “simultaneous conception and reduction to practice.” Id. This doctrine applies in circumstances where synthesis of the compounds is unpredictable. Accordingly, in our example, if synthesis of the subject chemical compounds is unpredictable, it could be argued that the scientist’s act of successfully synthesizing the compounds in the lab demonstrate that the scientist significantly contributed to the claimed invention and is therefore a proper inventor.
Key Considerations for Drug Developers that Utilize AI
In summary, companies should work with their patent counsel to:
- Evaluate existing patent portfolios claiming inventions developed at least in part through the utilization of AI. This includes an evaluation of the level of human involvement in the context of the Pannu factors.
- Assess existing AI-driven development programs and inventor evaluation processes. This should include a review and potential revision of IP policies and documentation of human involvement in AI development;
- Assess agreements with AI vendors for inventorship issues. This would involve evaluation of IP ownership provisions, as well as clauses that require proper natural persons to be names as inventors including vendor personnel; and
- Develop clear strategies for (1), (2) and (3) to strengthen inventorship positions and mitigate against validity challenges.
Expect also that inventorship challenges to AI-developed inventions will increase in the U.S. federal courts. The open question in the law and fact-specific standards provides a ripe ground for invalidity challenges in the federal courts. Companies should monitor relevant district court and (especially) Federal Circuit rulings to determine whether a standard for AI-assisted inventorship standards are established.
Conclusions
The law regarding inventorship for AI-driven inventions is uncertain. While helpful for framing the issues in the context of existing inventorship law, the USPTO’s Guidance confirms there is currently no bright line legal standard for AI-assisted inventions which leaves many open questions. Biotech innovators that utilize AI need to understand this, monitor forthcoming case law, and work with their patent counsel to develop a strategy to ensure that inventions developed through utilization of AI are protected.
Exhibit A
USPTO “Guiding Principles”
