Law360 (May 5, 2026, 6:19 PM EDT) -- Book publishers and legal novelist Scott Turow hit Meta Platforms Inc. with a proposed class action in New York federal court on Tuesday, accusing it of training its Llama large language models on millions of copyrighted books and articles from pirate sites instead of licensing the material.

The complaint came a day after textbook publisher Cognella Inc. filed separate lawsuits against Meta and Anthropic PBC in California federal court, alleging similar copyright violations tied to the companies' development of their LLMs.

The three suits frame the copyright infringement alleged as a byproduct of what they describe as an "arms race," accusing Meta and Anthropic of cutting legal corners as they rushed to keep pace with rivals in the generative text market. The complaints allege the competitive pressure drove the companies to obtain huge amounts of copyrighted material from pirate sites regardless of the legal risk.

"Defendants did not create the high-quality texts used in Llama's training datasets. Nor do defendants own or control the copyrights to those works," the publishers and Turow say in the New York complaint, which includes Meta CEO Mark Zuckerberg as a defendant. "Instead, Meta — at Zuckerberg's direction — copied millions of books, journal articles, and other written works without authorization, including those owned or controlled by plaintiffs and the class, and then made additional copies of those works to train Llama."

The complaints all have claims of direct copyright infringement via torrenting, alleging Meta and Anthropic copied and redistributed pirated works through peer-to-peer networks in the course of building their training datasets. Across the three complaints, the plaintiffs advance largely similar claims, also accusing the companies of contributory copyright infringement and violating the Digital Millennium Copyright Act by removing copyright information from materials to conceal their use.

Meta said in a statement to Law360 on Tuesday that it would "aggressively" fight the allegations.

"AI is powering transformative innovations, productivity and creativity for individuals and companies, and courts have rightly found that training AI on copyrighted material can qualify as fair use," the company said.

Anthropic did not immediately respond to a request for comment.

Meta, Anthropic and other leading LLM developers are facing complaints nationally that raise similar allegations about whether using copyrighted material for training is lawful.

Cognella says in its complaints that it chose to pursue its own suits rather than joining a proposed class action "to retain full control of its [cases]" and avoid "sprawling class-action settlements structured to resolve claims for pennies on the dollar."

The three complaints detail how Meta and Anthropic went to well-known piracy repositories such as Library Genesis, Z-Library and Anna's Archive.

"Adding insult to all of this injury, Anthropic's use of plaintiff's copyrighted materials facilitated Anthropic's creation of AI models capable of generating content that directly competes with and will compete directly with plaintiff's content," Cognella says. Its complaint against Meta includes a nearly identical line.

The publishers' suit in New York also alleges the tech companies' LLMs are competing directly with the works they have been trained on, with Llama able to immediately produce "sequels, prequels, spin-offs, and other adaptations" of those books.

"For example, prompted for a sequel to Scott Turow's bestselling novel, Innocent, Llama spits out a nearly 5,000 word, 10-chapter sequel, featuring Turow's iconic fictional characters (Rusty Sabich, Sandy Stern, Tommy Molto) and setting (Kindle County), while echoing Turow's signature writing style and creative choices by drawing on elements from the original," the complaint says.

In addition to damages, the plaintiffs are seeking injunctive relief that could bar Meta and Anthropic from continuing to use copyrighted works in training.

The book publishers and Turow are represented by Matthew J. Oppenheim, Jeffrey M. Gould, Daryl L. Kleiman, Edward Crouse and Eli Goldman of Oppenheim & Zebrak LLP, Megan K. Bannigan, James J. Pastore, Morgan A. Davis, Kathryn C. Saba and Abigail E. Liles of Debevoise & Plimpton LLP and Derek W. Loeser, Benjamin Gould, Chris N. Ryder and Samuel L. Rubinstein of Keller Rohrback.

Cognella Inc. is represented in both California complaints by Elizabeth Brannen, John Stokes, Lauren Martin, Christopher M. Rigali and Jacqueline Sahlberg of Stris & Maher LLP and Kyle Roche, Devin Freedman and Alex Potter of Freedman Normand Friedland LLP.

Counsel information for the defendants was not immediately available.

The cases are Elsevier Inc. et al. v. Meta Platforms Inc. et al., case number 1:26-cv-03689, in U.S. District Court for the Southern District of New York and Cognella Inc. v. Meta Platforms Inc., case number 3:26-cv-04053, and Cognella Inc. v. Anthropic PBC, case number 3:26-cv-04056, in U.S. District Court for the Northern District of California.

--Editing by Brian Baresch.

Elsevier is a division of RELX Group, which also owns Law360.

For a reprint of this article, please contact reprints@law360.com.