Five Major Publishers Sue Meta and Zuckerberg Over Alleged Illegal Use of Copyrighted Books to Train Llama

Meta trained Llama on large volumes of books and online text, while publishers and authors contend that copyrighted works cannot be included in datasets without authorization. Five major publishers and several authors have now filed a class-action lawsuit naming Meta CEO Mark Zuckerberg as a defendant. The case could influence how U.S. courts define “fair use” in generative AI training.

The plaintiffs' latest allegations say Meta illegally used millions of copyrighted books to train Llama and that Zuckerberg personally authorized the practice. As of July 20, 2026, the available information on the event did not identify the five publishers or specify the formal filing date or damages sought. The litigation is expected to focus on how Meta obtained the data, management's responsibility for the decision and whether its fair-use defense is valid.