The advent of artificial intelligence (AI), primarily through the ChatGPT software developed by OpenAI Inc., has taken the legal profession by storm.

Lawyers (and some nonlawyers) have relied on “legal research” performed by ChatGPT to their peril. These attorneys appropriate AI-generated research as their own, incorporating it into legal filings without verifying its accuracy. 

Unbeknownst to these lawyers, however, some of the cases provided by ChatGPT are fictitious — the result of the software “hallucinating.” Upon discovering the use of nonexistent cases, courts sanctioned these lawyers and referred them to licensing authorities for ethics violations. 

“A fake opinion is not ‘existing law’ and citation to a fake opinion does not provide a non-frivolous ground for extending, modifying, or reversing existing law, or for establishing new law. An attempt to persuade a court or oppose an adversary by relying on fake opinions is an abuse of the adversary system.” Park v. Kim, Docket No. 22-2057, slip op. at 10 (2d Cir. Jan. 30, 2024) (quoting Mata v. Avianca, Inc., No. 22CV01461 (PKC), 2023 WL 4114965, at *12 (S.D.N.Y. June 22, 2023)).

Impact on Design Professionals

However, the results of ChatGPT and other generative AI engines are not always fake. The results may be so authentic that they trigger possible copyright violations. This should concern every design professional — architects, engineers, etc. — who prepares plans and specifications and other instruments of service capable of being produced (or reproduced) through AI programs such as ChatGPT. 

An excellent example of why design professionals should be concerned lies in a recent decision in a copyright case involving actress and comedian Sarah Silverman. See Tremblay v. OpenAI, Inc., No. 23-cv-03223-AMO, slip op. (N.D. Cal. Feb. 12, 2024).

Before discussing the Tremblay case, we must review some key terms related to AI and some fundamental concepts related to copyright law.

AI Definitions

“‘Generative AI’ is a specific type of AI that uses computer algorithms to generate text, images, and other content based on the date on which it was trained. … Generative AI uses a process called deep learning, which uses artificial neural networks with multiple layers of processing to extract progressively higher-level features from data.” Practical Law Intellectual Property & Technology, “IT Basics: Generative AI and Large Language Models: Overview” (Thomson Reuters 2024) (IT Basics).

“A ‘Large Language Model’ (LLM) is a computerized language model that consists of an artificial neural network with a vast number of parameters. … The neural network’s deep learning algorithms consider[] the vast set of parameters to process and generate natural language text in a seemingly human manner.” See IT Basics.

Finally, a significant limitation of generative AI is its tendency to hallucinate. “Hallucination happens because LLMs are not capable of determining truthfulness or accuracy. Instead, they only predict what word ordering has the highest probability of responding to a prompt successfully.” See IT Basics.

Copyright Basics

Keep in mind these terms as we briefly examine the fundamentals of copyright law. The Copyright Act of 1976, a revision of the original statute passed in 1909, offers some key protections for design professionals. Specifically, 17 U.S.C. § 102(5) provides that diagrams, models and technical drawings are work product that may be copyrightable. 

In addition, 17 U.S.C. § 102(a)(8) offers copyright protection to a building design. For example, having a copyright for architectural work allows architects to maintain exclusive rights to their designs. They may reproduce, distribute, display and prepare work based upon their protected designs. These rights last for the lifetime of the author, plus 70 years. 

Formal copyright registration is not required for some protection but is recommended for more complete protection. The latter includes the ability to receive statutory damages, actual damages, lost profits, attorney’s fees (of the enforcing party) and possibly injunctive relief. 

Now that we have these basic definitions, we can ask the million-dollar question: Could the use of generative AI result in the infringement of an existing copyright? According to the Tremblay case, the answer is maybe.

The Tremblay Case

In Tremblay, Silverman and other authors filed suit against OpenAI over its alleged use of their books to train the LLM underlying the ChatGPT chatbot. By order dated Feb. 12, 2024, a federal judge in California dismissed parts of the lawsuit, rejecting the contention that the content generated by ChatGPT infringes the authors’ copyrights. 

Among other things, the makers of ChatGPT argued that their AI training is protected by the copyright doctrine (or defense) of fair use.

The judge dismissed parts of the complaint because the plaintiffs failed to prove “substantial similarity” between their copyrighted works and the ChatGPT output. 

“Distinctly, Plaintiffs here have not alleged that the ChatGPT outputs contain direct copies of the copyrighted books. Because they fail to allege direct copying, they must show a substantial similarity between the outputs and the copyrighted materials.” Tremblay, slip op. at 5. 

The court allowed the plaintiffs to file an amended complaint, so Silverman and her fellow authors may prevail at the end of the day — but the path is not easy. The Tremblay case is not the only example of failed copyright claims against AI companies.

Takeaways

At this stage of the game, the Tremblay case provides a cautionary tale for design professionals. First, design professionals should register their plans and specifications with the U.S. Copyright Office. This may be achieved through the Electronic Copyright Registration System. 

Second, consult an intellectual property lawyer if you learn that your instruments of service have been appropriated by another party (through the use of generative AI) without your authorization. The law of copyright is arcane, filled with twists and turns regarding claims and defenses. Proper legal counsel is critical. 

While this column provides a broad overview of a potential issue facing design professionals dealing with AI, it does little more than identify the issue through a current case (which, as of this writing, is still pending). A knowledgeable intellectual property lawyer is needed to navigate the AI minefield and assist design professionals in protecting their copyrighted instruments of service.

Disclaimer

 This article is for informational purposes only and not for the purpose of providing legal advice. Nothing in this article should be considered legal advice or an offer to perform services. The application and impact of laws may vary widely based on the specific facts involved. Do not act upon any information provided in this article, including choosing an attorney, without independent investigation or legal representation. The opinions expressed in this article are the opinions of the individual author and may not reflect the opinions of his firm.