As a proposed class action in California federal court that includes plaintiffs Sarah Silverman, Ta-Nehisi Coates and Junot Diaz reaches the summary judgment stage, six groups filed amicus briefs in support of each side's arguments by Friday's deadline.
The briefs highlight two emerging views in the case. One is that AI — and, in particular, generative AI, such as Meta's large language model Llama — is such a transformative technology that using copyrighted material to make it possible is protected by fair use. The other view is that generative AI exists today because it relies heavily on human creativity and expression, which should be compensated, and that Meta's actions cannot be fair use because its product competes with the original works.
"Defendant Meta Platforms Inc., a company valued at over a trillion dollars, asks this court to declare that it is free to appropriate and commercially exploit the content of copyrighted works on a massive scale without permission or payment for that content, a ruling that would have catastrophic consequences for authors and publishers of books, journals and other textual works protected by copyright," reads the brief from the Association of American Publishers, one of four groups that filed arguments in support of the plaintiffs.
However, four intellectual property law professors who filed a brief against the plaintiffs' summary judgment arguments said the copying the company has engaged in to train Llama is consistent with fair use because it is intermediate and internal, meaning it was done solely to develop the system.
"The case law, including binding circuit precedent, holds that internal copying, made in the course of creating new knowledge, is a transformative use that is heavily favored by fair use doctrine," said the professors from schools including Emory University and Harvard Law School. They said they "have sharply different views on the social costs and benefits of generative AI" but agree that "copyright is not the correct tool for managing those costs and benefits, especially with respect to training data."
In a statement to Law360 Monday, Meta called its large language model "transformational."
"Meta has developed transformational open source AI models that are powering incredible innovation, productivity and creativity for individuals and companies," Meta said. "Fair use of copyrighted materials is vital to this. We disagree with plaintiffs' assertions, and the full record tells a different story. We will continue to vigorously defend ourselves and to protect the development of GenAI for the benefit of all."
The Warhol Decision
The U.S. Supreme Court's 2023 decision in Andy Warhol Foundation for the Visual Arts v. Goldsmith
— the justices' most recent foray into the fair use analysis — is expected to influence the outcome of the litigation against Meta and other AI developers facing copyright complaints by content creators, including OpenAI, Microsoft and Anthropic.In Warhol, justices held that a silkscreen of music icon Prince that Warhol created based on a photograph by Lynn Goldsmith was not fair use, regardless of how transformative, because both works shared the same commercial purpose of magazine publishing. The amicus briefs in support of authors leaned into the decision in the case more than those in favor of Meta.
The Copyright Alliance, which represents the interests of thousands of individual creators and organizations, argued that Meta "improperly" wants the court to disregard the Warhol decision, "which clarified that even when a court finds a use to be transformative, that finding should have a limited effect on the ultimate fair use determination."
"The Supreme Court made clear in Warhol that whether a use is transformative is not dispositive of the question of fair use; rather, it is merely one subfactor within the first statutory fair use factor," the Copyright Alliance said. "Under Warhol, it is thus inappropriate for courts to attribute dispositive weight to what is simply one aspect of the first-factor analysis," which also includes the question of whether the use is commercial.
"Llama is clearly part of a commercial venture, designed to attract as many users as possible and bolster Meta's position in the market," the Copyright Alliance said.
The second fair use factor looks at the type of work that was copied and the third at the amount that was taken. The fourth and final factor looks at the potential impact of the copying on the market for the original.
Torrenting and Fair Use
The IP professors who filed a brief opposing the plaintiffs' arguments did not mention the Warhol case.
Meanwhile, the Electronic Frontier Foundation, which describes itself as a defender of "civil liberties in the digital world," focused as much on what the justices said in Warhol as what they did not say.
Quoting from the Warhol decision, the foundation said fair use "permits courts to avoid rigid application of the copyright statute when, on occasion, it would stifle the very creativity which that law is designed to foster." It then mentioned what Warhol did not say as it urged the court not to be swayed by the plaintiffs' attempt to "shortcut" the fair use analysis by focusing on the "legally tangential issue" of how Meta acquired the works to train Llama.
That comment is a reference to evidence the plaintiffs obtained through discovery indicating that Meta downloaded millions of pirated copyrighted works from peer-to-peer networks — a practice known as torrenting — to train its AI model.
The Supreme Court's "most recent fair use opinion ignores bad faith altogether in its analysis of the first statutory factor," the Electronic Frontier Foundation said in arguing that Meta's use of those pirated works should not affect its fair use defense.
"First, there is no requirement that a would-be fair user must possess an authorized copy of a work," the foundation said. "Fair use involves an overall balancing of the equities, including the potential public benefits of the use."
The group argued that even if "good faith" was a relevant factor in the fair use analysis, using torrenting instead of other means to obtain books to train an AI model does not show a lack of good faith.
Meta's torrenting, however, was the main thrust of the amicus brief in support of the plaintiffs by the International Association of Scientific, Technical and Medical Publishers, which argued that the fair use doctrine is not meant to legitimize "commercial-scale copying" of content from illegal sources.
"The implications of Meta's logic are astounding — anything goes in sourcing copyrighted content as long as the final end use is supposedly non-infringing," the group said. "As long as you're 'just learning from it,' why ever buy a copyrighted work? Instead, go to a piracy hot spot and copy at will."
Ten IP professors who filed a brief in support of the plaintiffs called Meta's fair use arguments "breathtaking" and said the company was asking "for greater legal privileges than courts have ever granted human authors." The professors said training generative AI models with works from human authors is not transformative.
The professors, whose schools include Columbia Law School, the George Washington University Law School and the University of Texas at Austin, said a ruling favoring Meta would undermine the rights of creators everywhere.
"Since not only Meta, but OpenAI, Google, Anthropic, xAI, and many others are creating generative AI models through unauthorized use of works protected by copyright, a decision in Meta's favor here will have the cascading effect of undermining the rights of copyright holders in every domain rapidly, with ease, and at a scale unimaginable just a few years ago," the professors said.
Counsel for the plaintiffs thanked the groups that filed briefs on their behalf.
"We appreciate the broad support for protecting the intellectual property of millions of authors and creators in this country. As the International Association of Scientific, Technical, and Medical Publishers point out in support of Plaintiffs' summary judgment motion, 'Meta knowingly copied and distributed a shocking amount of infringing material from the world's most notorious infringing websites,' making a 'calculated risk to put its own competitive interests ahead of millions of copyright holders' rights,'" Maxwell Pritt of Boies Schiller Flexner LLP said in a statement to Law360.
Meta's counsel didn't immediately respond to a request for comment Monday
The authors are represented by Joseph R. Saveri, Cadio Zirpoli, Christopher K.L. Young, Holden Benon and Aaron Cera of Joseph Saveri Law Firm LLP, David Boies, Joshua Irwin Schiller, Maxwell V. Pritt, Jay Schuffenhauer, Joshua M. Stein, Margaux Poueymirou, David L. Simons and Jesse Panuccio of Boies Schiller Flexner LLP, Elizabeth J. Cabraser, Daniel M. Hutchinson, Reilly T. Stoler, Rachel Geman, Kenneth S. Byrd and Betsy A. Sugar of Lieff Cabraser Heimann & Bernstein LLP, Bryan L. Clobes, Mohammed A. Rathur and Alexander J. Sweatman of Cafferty Clobes Meriwether & Sprengel LLP, Nada Djordjevic, Amy E. Keller, David A. Straite and James A. Ulwick of DiCello Levitt LLP, Scott J. Sholder and CeCe M. Cole of Cowan Debaets Abrahams & Sheppard LLP and Matthew Butterick.
Meta is represented by Bobby Ghajar, Mark Weinstein, Kathleen Hartnett, Judd Lauter, Elizabeth L. Stameshkin and Colette Ghazarian of Cooley LLP and Angela L. Dunning of Cleary Gottlieb Steen & Hamilton LLP.
The Electronic Frontier Foundation is represented in-house by Mitchell L. Stoltz.
The four IP law professors are represented by Rebecca Tushnet of Harvard Law School and Chris K. Ridder of Ridder Costa & Johnstone LLP.
The Copyright Alliance is represented in-house by Keith Kupferschmid and Kevin R. Madigan and by Benjamin S. Akley, Frank P. Scibilia and Joshua Weigensberg of Pryor Cashman LLP.
The Association of American Publishers is represented by Jacqueline C. Charlesworth, Ruby A. Strassman and Kristen G. Niven of Frankfurt Kurnit Klein & Selz PC.
The International Association of Scientific, Technical and Medical Publishers is represented by Thomas A. Harvey of Coblentz Patch Duffy & Bass LLP and Michele H. Murphy of Oppenheim & Zebrak LLP.
The 10 IP law professors are represented by Sabita J. Soneji of Tycko & Zavareei LLP.
The case is Richard Kadrey et al. v. Meta Platforms Inc. et al., case number 3:23-cv-03417, in the U.S. District Court for the Northern District of California.
--Additional reporting by Dorothy Atkins and Hailey Konnath. Editing by Karin Roberts.
For a reprint of this article, please contact reprints@law360.com.
