Publishers sue Google over Gemini training data

Hachette Book Group, Cengage Learning, Elsevier, Scott Turow and S.C.R.I.B.E. have sued Google in the US District Court for the Southern District of New York, accusing the company of using copyrighted books and journal articles to train Gemini without permission or payment. The complaint filed on 10 July seeks class-action status and calls the alleged copying “one of the most prolific infringements of copyrighted materials in history.”

The plaintiffs say Google repurposed works collected for Google Books, Google Play Books and Google Scholar, where permissions were limited to search snippets or ebook sales. The suit also alleges Google relied on web scrapes from “known pirate sources” and paywalled material, and says an internal document warned that book-based training could be “highly problematic” and create “$10Bs-$100Bs in potential fines.”

The publishers argue Gemini can compete with the works used for training, producing near-verbatim passages, textbook-like chapters and knockoff fiction. The complaint says Gemini can create a 100-page murder mystery in about 20 minutes for 39 cents, and asks for statutory damages, a permanent injunction and destruction of unauthorised training copies. Google has not commented.

The case adds to copyright litigation against OpenAI, Anthropic and Meta. The same publishers sued Meta earlier this year, while Anthropic separately agreed to pay $1.5bn to authors over pirated copies.