By: Shashank Upadhye
Introduction
Artificial Intelligence has revolutionized industries ranging from pharmaceuticals to creative arts, raising profound questions about intellectual property (IP) protection. The intersection of AI and IP law presents challenges in patents, copyrights, trade secrets, and broader policy considerations. This article explores the evolving landscape of AI-related IP, with a focus on patents and copyrights, and examines future considerations such as AI’s role in patent law, including whether AI can be deemed a “person of ordinary skill in the art.”
Patent Issues in AI
Inventorship and AI-Generated Inventions
One of the most debated questions in AI patent law is whether an AI system can be named as an inventor. Most jurisdictions, including the U.S. Patent and Trademark Office (USPTO) and the European Patent Office (EPO), have ruled that only humans can be inventors. The case Thaler v. Vidal reaffirmed that AI systems, such as the DABUS AI, cannot be considered inventors under U.S. patent law. The ruling emphasized that patents require “natural persons” as inventors, leading to ongoing discussions about whether AI-generated inventions can be protected at all. Thaler v. Vidal, 43 F.4th 1207, 1213 (Fed. Cir. 2022).
AI-Assisted Inventions and the Pannu Factor Test
While AI cannot be an inventor, the role of AI-assisted inventions is still evolving. Courts rely on the Pannu v. Iolab Corp. test to determine human contribution to an AI-assisted invention. Pannu v. Iolab Corp., 155 F.3d 1344, 1351 (Fed. Cir. 1998).
The test considers whether a human:
- Made a significant intellectual contribution to the invention.
- Contributed in a way that is not insignificant compared to the overall invention.
- Did more than merely explain well-known concepts or existing knowledge.
This test is crucial in determining inventorship when AI tools play a significant role in generating patentable innovations. That is, whilst AI is machine based, significant human involvement is needed.
AI and the Person of Ordinary Skill in the Art (POSITA)
The POSITA standard is fundamental to patent law because it underlies obviousness (§103) and enablement (§112) under U.S. patent law. The mere patent law phrase “person” of ordinary skill might betray the role of AI, because AI is not a person.
In U.S. patent law, a POSITA is a hypothetical person who has, among other things: (i) ordinary skill and knowledge in a specific technical field; (ii) access to publicly available prior art; and (iii) the ability to apply common sense and routine problem-solving techniques. See, Shashank Upadhye, Generic Pharmaceutical Patent and FDA Law, section 2.5 (2024-2025 Ed.)(available here: https://tinyurl.com/yd2353sc). See also, Daiichi Sankyo Co., Ltd. v. Apotex, Inc., 501 F.3d 1254, 1256 (Fed. Cir. 2007)(“Factors that may be considered in determining level of ordinary skill in the art include: (1) the educational level of the inventor; (2) type of problems encountered in the art; (3) prior art solutions to those problems; (4) rapidity with which innovations are made; (5) sophistication of the technology; and (6) educational level of active workers in the field.” Envtl. Designs, Ltd. v. Union Oil Co., 713 F.2d 693, 696 (Fed. Cir. 1983)).
Issues with AI as a POSITA
If AI were recognized as a POSITA, it could have major implications for patent law. Several key issues arise. First is the non-human nature of AI. Patent law assumes a human perspective when evaluating inventions. AI, as a machine, lacks human intuition, creativity, and reasoning. The POSITA involves “ordinary creativity” (can AI have “ordinary” creativity) or does it follow a fundamentally different mode of problem-solving? AI does not “think” in the same way as humans; it generates solutions based on pattern recognition and probabilistic modeling.
How AI Impacts Obviousness Analysis
If AI is considered a POSITA, it may increase the likelihood of finding inventions obvious because AI can rapidly generate ideas and solutions that humans would not have readily conceived. AI-assisted innovation may accelerate technological progress, making prior art more accessible and comprehensive. Thus, as the universe of prior art becomes more readily available, then obviousness rejections during prosecution or invalidity challenges increase. Further, Courts and patent offices may struggle to determine what is “obvious” to an AI versus what is obvious to a skilled human.
Enablement & AI’s Role
Enablement requires that a patent teaches a POSITA how to make and use an invention without undue experimentation. Baxalta Incorporated v. Genentech, Inc., 81 F.4th 1362, 1365 (Fed. Cir. 2023). If AI is a POSITA, does it change what is considered “undue experimentation”? AI can analyze vast amounts of data quickly, which may shift the enablement standard. That is, AI could make undue experimentation to a human more routine for a machine. This could lead to more enablement rejections or challenges.
What’s Next for AI and the POSITA
The first option is to maintain the status quo. Therefore, courts, legislatures, and patent offices maintain that AI is a tool, but not a POSITA, and continue assessing patents based on human capabilities. A second option is the hybrid approach. This considers AI as an augmenting tool but not as a full replacement for human skilled artisans. The third approach is the fully new paradigm that recognizes AI as a POSITA but develop guidelines for determining how AI-generated insights affect patentability.
Future AI Issues in Patent Law
Should AI Be a Co-Inventor?
Because of the existing rules on human intervention, changes to AI inventorship would require legislative change or policy changes. Regulators tend not to act in significant ways until the need because more dire.
Disclosure & Candor Requirements for AI in Patent Applications
Another emerging issue is whether inventors should disclose the use of AI in the patent application process. USPTO rules require a duty of candor and duty of disclosure. When AI is used to draft the application, it necessarily relies on the prior art for the drafting. But AI outputs rarely provide the citation where the information came from. Will the lack of an actual citation to the prior art mean that the applicant breached the duty to disclose relevant and material prior art? Further, if AI drafts the application and characterizes the prior art, must the applicant check the characterization to ensure that the art is not mischaracterized by the AI? There are likely other AI related issues that may arise in the future.
Copyright Issues in AI
Authorship and Ownership of AI-Generated Works
The U.S. Copyright Office has ruled that AI-generated works are not copyrightable unless they have sufficient human input. In Thaler’s Creativity Machine case (2022), the Copyright Office denied copyright protection for an artwork autonomously created by AI. Courts have consistently maintained that copyright law applies only to human authors. Thaler v. Perlmutter, 687 F.Supp.3d 140, 147 (D.D.C., 2023)(“The understanding that “authorship” is synonymous with human creation has persisted even as the copyright law has otherwise evolved.”).
However, determining authorship in AI-assisted works is complex. If a human provides substantial creative input (e.g., by guiding AI-generated text, music, or images), they may be considered the author. This raises questions about how much human intervention is needed for copyright protection.
Fair Use and AI Training Data
AI models often rely on copyrighted data for training, leading to significant legal challenges. Under Section 107 of the U.S. Copyright Act (17 U.S.C. §107), courts assess fair use based on a case-by-case analysis that considers:
- Purpose and character of use (e.g., whether AI training is transformative or commercial).
- Nature of the copyrighted work (e.g., factual vs. highly creative content).
- Amount and substantiality of the portion used (e.g., whether AI uses entire works or only extracts elements).
- Effect on the market (e.g., whether AI-generated content competes with the original work).
See, Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569, 577 (1994). Lawsuits such as Getty Images v. Stability AI challenge whether AI training on copyrighted images constitutes fair use. If courts rule against AI companies, they may need to license data before training AI models. See, Thomson Reuters Enterprise Centre GMBH v. Ross Intelligence Inc., 2025 WL 458520, at *1 (D. Del. 2025)(involving the training of legal research database using another’s headnote synopses).
Trade Secrets – You Can’t Claw It Back
Trade secret protection is valuable, but its value intrinsically derives from its secrecy. And AI’s intrinsic value is its ability to cull the world’s public information. What happens when trade secrets are uploaded into the AI world? For example, suppose a disgruntled employee uploads a company’s trade secrets into the AI? Of course, the company can sue the employee for the misappropriation of the secrets, but what remedy can be obtained? Once the secrets are uploaded into the AI, they can’t be clawed back. The company cannot compel the AI to return the secret information and block future access to any secrets. This is why it’s important to have good governance in place to control trade secret access.
Best Practices
Just the like the underlying information in the AI is constantly evolving, so to must best practices. For AI and patents/copyrights, one should develop a clear AI IP strategy, including as much “human intervention” as you can to support patents and copyrights. And such strategies should be documented.
Further always consider utilizing multiple forms of protection – patents, copyrights, trade secrets. We also suggest having robust contracts for employer ownership of IP, assignments, trade secrets protection, but also policies on how to use AI properly. In other words, have good corporate governance models in place. And finally, you should monitor emerging legal developments and compliance requirements to ensure you are current.
How we can help you?
Our firm specializes in patent strategy and litigation. We help clients analyze the portfolio of patents. For investors, we provide the deep dive due diligence to find any problems in the patent portfolio or vet out the accuracy of pitch-decks. We help clients in patent litigation, appeals, counseling, opinions of counsel, and PTAB proceedings. When your current firm needs help or the client needs a change of counsel, we can help.
About Upadhye Tang LLP
Upadhye Tang LLP is an IP and FDA boutique firm concentrating on the pharmaceutical, life sciences, and medical device spaces. We help clients with navigating the legal landscape by helping on counseling and litigation. Clients call us to help move drug and device approvals along and to represent them in IP and commercial litigation. Call Shashank Upadhye, 312-327-3326, or by email: shashank@ipfdalaw.com, for more information.
No Attorney-Client Relationship or Legal Advice
The content of this website has been prepared by Upadhye Tang LLP for informational purposes only and does not constitute legal advice. Nothing shall not be construed as an offer to represent you, nor is it intended to create, nor shall the receipt of such information constitute, an attorney-client relationship. Please call us first with any questions about the firm.
Nobody should take any action or choose not to take any action based upon the information on this website without first seeking appropriate professional counsel from an attorney licensed in the user’s jurisdiction. Because the content of any unsolicited Internet email sent to Upadhye Tang LLP at any of the email addresses set forth in this website will not create an attorney-client relationship and will not be treated as confidential, you should not send us information until you speak to one of our lawyers and get authorization to send that information to us.
Individual Opinions
The opinions and views expressed on or through this website, or any newsletter, are the opinions of the designated authors only, and do not reflect the opinions or views of any of their clients or law firms or the opinions or views of any other individual. And nothing said in any newsletter or website reflects any binding admission or authority.
