Meta hit with new copyright lawsuit over alleged use of pirated books to train Llama AI

A fresh legal fight is brewing over how AI models are built, and this time it puts Meta back in the spotlight. A group of major publishers, including Elsevier, Cengage, Hachette, Macmillan, and McGraw-Hill, has sued Meta Platforms in Manhattan federal court, alleging the company used pirated books and journal articles to train its Llama AI models, according to a Reuters report.

The complaint alleges that the company copied millions of copyrighted works without permission and fed them into its systems to improve its responses to user prompts.

The case pulls in author Scott Turow and is framed as a proposed class action that could expand to include a wide range of rights holders. At its core, the lawsuit argues that Meta’s training practices cross a line by relying on material that was never licensed or authorized for this use, Reuters reported.

“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,” a Meta spokesperson said in response. “We will fight this lawsuit aggressively.”

Copyright clash escalates as pubishers sue Meta over alleged misuse of pirated content for AI training

The publishers paint a very different picture. They say the dataset behind Llama includes everything from textbooks and academic papers to well-known titles like The Fifth Season by N.K. Jemisin and The Wild Robot by Peter Brown. The complaint asks the court to certify the case as a class action on behalf of a broader class of copyright owners and to award damages, though the exact amount has not been specified.

“Meta’s mass-scale infringement isn’t public progress, and AI will never be properly ‌realized ⁠if tech companies prioritize pirate sites over scholarship and imagination,” said Maria Pallante, president of the Association of American Publishers.

This is not Meta’s first run-in with authors over AI training. In late 2023, a separate group that included comedian Sarah Silverman and novelist Michael Chabon accused the company of using their books without permission. Parts of that case were dismissed, though the plaintiffs were given the chance to revise their claims and move forward.

The new lawsuit adds pressure at a time when courts are still sorting out how copyright law applies to AI. Creators across publishing, news, and the arts have taken aim at companies such as Meta, OpenAI, and Anthropic, arguing that their work has been used to build commercial systems without their consent.

Judges have not landed on a single answer. Early rulings have pointed in different directions, leaving companies and creators without a clear rulebook. That uncertainty has already pushed at least one major player to settle. Anthropic, backed by Amazon and Google, agreed last year to pay $1.5 billion to resolve claims from a group of authors, closing a case that carried the risk of far more serious damages.

This latest filing signals that the fight is far from over. At stake is a question that could shape the future of AI: whether training models on copyrighted material constitutes fair use or amounts to infringement. The answer will decide how freely companies can build the next generation of AI systems and how much control creators retain over their work.