Publishers sue Google over alleged copyright theft for Gemini training

Publishers filed a putative class action in Manhattan on July 10 accusing Google of using millions of copyrighted books and journal articles to train Gemini without permission. The case, brought in U.S. District Court for the Southern District of New York by Hachette Book Group, Cengage Learning, Elsevier and author Scott Turow, seeks statutory damages and an injunction.

The complaint says Google copied material obtained for Google Books and other services, downloaded unauthorized web scrapes from across the internet, including from known pirate sources and behind paywalls, and then copied those works many times over to train its large language models. The plaintiffs also allege Google stripped copyright management information from the copied works and used a purpose-built system designed to generate content that directly substitutes for original books and articles.

Internal Google documents warned that using publisher-provided books for AI was "highly problematic for Google" and could expose the company to "$10Bs-$100Bs" in fines. Ten representative works allegedly copied without permission include Principles of Economics by M. Gregory Mankiw, Innocent by Scott Turow and The Fifth Season by N.K. Jemisin. Maria A. Pallante, president and chief executive of the Association of American Publishers, said the case could have "far-reaching consequences" for the legal rights, remedies and livelihoods of authors and publishers.

Hachette and Cengage had previously moved to intervene in the separate In Re Google Generative AI Copyright Litigation in Northern California before Judge Eumi K. Lee. They withdrew that motion to preserve claims that could fall outside the earlier proposed class and to avoid possible three-year statute-of-limitations problems. The publishers said direct participation was needed because Google objected to including publishers as class representatives.

In May 2026, nearly the same roster of publishers sued Meta over alleged use of pirated books to train Llama models. Anthropic agreed in 2025 to pay authors $1.5 billion to settle a class action over alleged piracy, while The New York Times’ suit against OpenAI and Microsoft remains ongoing.