Pirated copies for AI training: Publishers sue Meta

Five major publishers and bestselling author Scott Turow have filed a lawsuit against Meta and Mark Zuckerberg personally for ’one of the most massive infringements of copyrighted materials in history’. They describe the ’illegally torrented millions of copyrighted books and journal articles from notorious pirate sites’ for training the AI technology Llama. The lawsuit is supported, among others, by Holtzbrinck subsidiary Macmillan and the scientific publisher Elsevier. Together, they accuse Meta and Zuckerberg that the decision to infringe copyright was ‘deliberate’. Meta acted according to the motto ‘move fast, and break things’ and must now repair what has been broken.

Flood of AI-generated books as competition

In the class action lawsuit (Case No.: 1:26-cv-03689), which the Association of American Publishers made public, it is once again about AI training using collections of copyrighted content. In a previous proceeding, Meta had already admitted to downloading the content via BitTorrent from so-called shadow libraries. It was once defended by stating that the data had not been passed on – via “seeding”. In total, it concerns enormous amounts of data that were procured and used in an attempt to win the AI race. As a result, however, a technology has also emerged that is now being used to compete with publishers and the authors published there.

The plaintiffs further explain what this competitive situation looks like. For example, a user is reported who explains how Meta's Llama AI was prompted with a single request to write a 100-chapter fictional book. In addition, the person claimed that the AI technology could write entire scientific articles and textbooks. Another person published three books in three months and forgot to remove the prompt with which the responsible AI was instructed to adapt the writing style to that of an author: "These AI-generated books are already flooding the world’s largest book marketplace, Amazon, in volumes that materially displace human-authored works." It is already a real crisis.

For Meta, the lawsuit is not a novelty; the Facebook group has been the focus of attention for its actions in AI training for some time. In the summer, it prevailed in another proceeding against 13 renowned authors, but only thanks to the “half-hearted” argumentation of the plaintiffs. The legal victory did not explicitly mean that Meta's actions were lawful, the judge explained. The company has now announced to the New York Times that it intends to fight the new lawsuit “aggressively”: “A.I. is powering transformative innovations, productivity and creativity for individuals and companies” and courts have found that this type of training can be legal, claims a spokesperson.

(mho)