Creating Precedent: Generative AI Litigation Theories of Liability

By: Jack Mansur, Class of 2025

Background

Artificial intelligence (“AI”) describes how computers perform tasks generally viewed as requiring human intelligence.[1] As a branch of AI, generative artificial intelligence (“GAI”) utilizes models and algorithms to generate new information, such as images, code, and music.[2] Examples of software include Stability AI’s Stable Diffusion, Google’s Bard, and GitHub’s Copilot, amongst others.[3] While GAI has existed for years, in recent months the legal world has been introduced to new and emerging issues involving this advancing technology.[4]

OpenAI, arguably the most prominent of the current GAI research and development companies, is the designer of the well-known and ground-breaking chatbot, ChatGPT.[5] Experts believe OpenAI trained its model on hundreds of gigabytes of text data harvested from books, articles, and websites.[6] However, the exact sources and references of training data for the models and other GAI systems have been kept confidential.[7] As alleged, OpenAI and additional companies have “raced to release AI technology with disregard for the catastrophic risk to humanity in the name of ‘technological advancement.’”[8]

Legal Theories

Surging litigation challenging the creators and developers of this powerful technology has exposed potential issues with biased training data in building these models.[9] Complaints primarily fall within two broad buckets of liability theories: (1) intellectual property (“IP”) infringement and (2) invasion of privacy, property, and publicity theories.

At issue in ongoing IP litigation is whether GAI’s use of training data and incorporation of such data into the output constitutes an infringement of copyright owners’ rights and whether GAI companies have the legal right to use existing images, text, or code to train their algorithms.[10] For example, in a putative class action against GitHub, Microsoft, and OpenAI, Plaintiffs claim license violations and infringement of IP rights where GitHub and Microsoft allegedly stripped Plaintiffs’ “attribution, copyright notice, and license terms from their code in violation of . . . [their] rights.”[11]

Other lawsuits challenge GAI’s use of training data on invasion of privacy, property, and publicity rights. In a 151-page putative class action complaint against OpenAI and Microsoft, Plaintiffs allege invasion of privacy with an underlying theme of ensuring the future existence of humanity.[12] Plaintiffs assert AI companies now have enough information to “create our digital clones” and thus, in aggregate, could replicate our voice, predict our next move, or misappropriate our skill sets, obliterating privacy as we know it.[13]

Next Steps

The lack of precedent creates an open question of how far are courts willing to go to regulate GAI. Change is coming, and all eyes will be on the courts at the forefront. Society is well into the age of GAI technology, and while negative connotations have been a mainstay in news headlines through the surge of lawsuits, GAI has the potential to transform our society for the better. Surprisingly, many Plaintiffs realize this notion. Embedded in many of the complaints is the idea of a temporary pause. Plaintiffs want “adequate safeguards” that contribute to the “greater societal good by advancing human rights, promoting social justice, reducing inequities, and empowering marginalized groups.”[14] Ethically speaking, this message should hold for all individuals and companies, not just lawyers, to approach the powerful technology cautiously and optimistically as it transforms the way we do business and the world as we know it.

[1] See Catherine Casey, Ronald J. Hedges, Marissa J. Moran, Stephanie Wilson, Generative AI in Legal Practice: What It Is and How Lawyers Can Use It, The American Law Institute Continuing Legal Education (June 28, 2023), https://1.next.westlaw.com/Link/Document/Blob/Ie1a439d4205611ee9936ad00d20c6353.pdf?originationContext=document&transitionType=DocumentImage&uniqueId=f7ab18ad-caec-4bbf-837c-4bc8db609571&ppcid=409c4988d3ec4c38bab48b92c948f5db&contextData=(sc.Default). [2] See Sneha Kothari, Top Generative AI Tools: Boost Your Creativity, SimpliLearn (June 21, 2023), https://www.simplilearn.com/tutorials/artificial-intelligence-tutorial/top-generative-ai-tools. [3] See Konstantine Buhler, Generative AI Is Exploding. These Are The Most Important Trends You Need To Know, Forbes (Apr. 11, 2023, 6:15 AM), https://www.forbes.com/sites/konstantinebuhler/2023/04/11/ai-50-2023-generative-ai-trends/?sh=3c2485507c0e. [4] See Lisa T. Oratz, D. Sean West, Tim Carter, Tyler Robbins, A New Generation of Legal Issues Part 2: First Lawsuits Arrive Addressing Generative AI, Perkins Coie (Apr. 20, 2023), https://www.perkinscoie.com/en/news-insights/first-lawsuits-arrive-addressing-generative-ai.html. [5] See Compl. at 30, Plaintiffs P.M., et al. v. OpenAI LP, Microsoft Corp., et al., No. 3:23-cv-03199-JCS (N.D. Cal. June 28, 2023). [6] Id. [7] See Will Bedingfield, The Generative AI Battle Has a Fundamental Flaw, WIRED (July 25, 2023, 7:00 AM), https://www.wired.com/story/artificial-intelligence-copyright-law/?bxid=63e7e66eb1e35d1ac00b9037&cndid=72886146&esrc=BottomStories&mbid=m%E2%80%A6. [8] Compl. at 4, Plaintiffs P.M., et al. v. OpenAI LP, Microsoft Corp., et al., No. 3:23-cv-03199-JCS (N.D. Cal. June 28, 2023). [9] See Id. at 3 (claiming OpenAI trained products on stolen data without implementing proper safeguards or controls to prevent further legal violations or endangerment of lives). [10] See John Quinn, The Clash of Generative AI and Intellectual Property Law: What It Means For Businesses, Forbes (May 3, 2023, 9:00 AM), https://www.forbes.com/sites/forbesbusinesscouncil/2023/05/03/the-clash-of-generative-ai-and-intellectual-property-law-what-it-means-for-businesses/?sh=251204916c01. [11] Compl. at 3, Does v. GitHub, Inc., Microsoft Corp., OpenAI, Inc., et al., No. 4:22-cv-06823-JST (N.D. Cal. Nov. 3, 2022). [12] See Compl. at 3–4, Plaintiffs P.M., et al. v. OpenAI LP, Microsoft Corp., et al., No. 3:23-cv-03199-JCS (N.D. Cal. June 28, 2023). [13] See Id. at 7. [14] Id. at 59; see also Compl. at 54–56, J.L., et al. v. Alphabet, Inc., Google DeepMind, and Google L.L.C., No. 3:23-cv-3440-LB (N.D. Cal. July 11, 2023) (requesting establishment of AI Council, transparency protocols, opt-out requirements, monetary fund, amongst additional technological safety measures).