A Google Gemini copyright lawsuit filed by some of publishing’s largest names accuses the company of training its artificial intelligence models on books it had no right to use, and of then deliberately scrubbing the evidence. The case, brought in the U.S. District Court for the Southern District of New York, names Hachette, Cengage, Elsevier, bestselling thriller writer Scott Turow, and advocacy group S.C.R.I.B.E. as plaintiffs.
The Google Gemini Copyright Lawsuit and What It Alleges
The complaint describes Google’s conduct as The Guardian characterised it: ‘one of the most prolific infringements of copyrighted materials in history.’ The plaintiffs argue that Google used books made available through Google Books and Google Play Books for AI training purposes, well beyond the limited scope those programmes permitted. Publishers Weekly reports that the complaint also names Google Scholar, a research platform, as an additional alleged source of copied material.
Google Books was built, the suit acknowledges, on a long-standing arrangement: publishers and authors provided their copyrighted works so that users could search for books and read short snippets. The plaintiffs say Google took that access and went much further, copying the underlying texts to feed its Gemini models. ‘Google illegally copied works from all these scope-limited programs for AI training, knowing it lacked authorization to do so,’ the complaint states.
The suit also points to an internal company document that reportedly warned that using copyrighted books for AI training could be ‘highly problematic for Google’ and might result in ‘$10Bs-$100Bs in potential fines.’ Good e-Reader reports that the complaint further alleges Google employees preferred training on works by esteemed authors, and that executives understood publishers would regard some of the copying as illegal.
Beyond damages, the plaintiffs are seeking a permanent injunction to halt further infringement and an order compelling Google to destroy all unauthorised copies of their works. Google did not respond to requests for comment.
The Same Plaintiffs Have Done This Before
This is not the first time these particular publishers and authors have pursued this strategy. Hachette, Cengage, Elsevier, and Turow filed a near-identical suit against Meta in May 2026, challenging how that company’s Llama models were trained, also in the Southern District of New York. Elsevier publishes scientific journals including The Lancet and Cell; Turow is the author of Presumed Innocent and Presumed Guilty.
The Google action also follows an earlier, unsuccessful attempt by Hachette and Cengage to join the Google Generative AI Copyright Litigation, a separate case brought by illustrators and writers in 2023 that Google had been contesting the publishers’ right to join. Filing fresh proceedings gives the publishers a cleaner legal footing and, critically, a different judge.
That jurisdictional choice carries real weight. Two federal judges in California have already ruled that training AI on copyrighted works constitutes fair use under U.S. copyright law. Those decisions do not bind courts in New York, and the publishers will be hoping a Southern District judge takes a different view of the same arguments.
What the Anthropic Settlement Means for the Industry
The most instructive precedent so far has arrived not from a courtroom verdict but from a settlement. Anthropic agreed to pay $1.5 billion in the Bartz v. Anthropic case, the largest copyright payout in U.S. legal history, after court documents showed it had downloaded over 7 million books from pirate sites, including LibGen and Pirate Library Mirror, to use as training data.
Judge William Alsup of the U.S. District Court for the Northern District of California had drawn a careful line in June 2025. Training on books acquired legitimately, he found, could qualify as fair use. Maintaining a ‘central library’ built on pirated copies could not. The Authors Guild has advised members that the settlement also requires Anthropic to destroy the unlawfully obtained files.
Around half a million writers were eligible for payments of at least $3,000 under the settlement, though many opted out to pursue independent litigation. Approximately a quarter of the $1.5 billion fund is earmarked for legal fees, according to the Kluwer Copyright Blog. Details of qualifying criteria, including ISBN or ASIN registration requirements, are set out by Lieff Cabraser, lead counsel for the class.
The broader litigation landscape keeps widening. CNN filed suit against AI company Perplexity in May 2026, alleging it illegally copied more than 17,000 CNN stories. As of mid-July 2026, 17 news organisations including The New York Times were pressing OpenAI to disclose training evidence in the copyright case that newspaper originally filed in 2023.
For Google, the Southern District of New York now becomes the arena that matters most. Whether a judge there follows California’s fair-use reasoning, or treats the allegation that copyright information was deliberately removed as a different kind of wrong altogether, will shape how much legal exposure every AI developer is carrying.
