Google sued by major publishers over Gemini AI training

On May 17, 2026, a major copyright dispute escalated when a coalition of publishers and authors launched a class action lawsuit against Google in U.S. District Court for the Southern District of New York. The plaintiffs—Hachette, Cengage, Elsevier, bestselling author Scott Turow, and the author advocacy group S.C.R.I.B.E.—accuse Google of using their copyrighted works without permission to train Gemini, the company's flagship generative AI assistant.

The lawsuit's core claim is twofold: first, that Google misappropriated copyrighted material; second, that the company deliberately altered or removed copyright notices to hide the fact that Gemini was trained on what the plaintiffs characterize as "stolen materials." Critically, the plaintiffs point to Google Books as evidence of a deliberate misuse of trust. In that program, publishers and authors had provided Google with access to their works for a narrow, specific purpose: making books searchable through snippet-based results that allow users to see bibliographic information and brief excerpts, but not entire texts. According to the lawsuit, Google then took copies of those same books—as well as books uploaded to the Google Play store—and fed them into Gemini's training pipeline without ever requesting or obtaining authorization.

The complaint quotes the lawsuit directly: "Google illegally copied works from all these scope-limited programs for AI training, knowing it lacked authorization to do so." More damaging still, the plaintiffs cite an internal Google document that allegedly acknowledges the company's own legal risk. That memo states that using copyrighted books for AI training could be "highly problematic for Google" and might trigger "$10Bs-$100Bs in potential fines." The fact that Google's own analysis flagged the practice as both legally and financially hazardous suggests the company proceeded with knowledge of the jeopardy it was creating.

This lawsuit is one of many such claims. Publishers, authors, and copyright holders have filed complaints against Google, Meta, OpenAI, and Anthropic over similar allegations. To date, two California courts have ruled that the use of copyrighted works for AI training qualifies as "fair use" under U.S. copyright law—a doctrine designed long before the internet and AI existed. Those rulings have been favorable to the AI companies, but the law remains contested. Anthropic, one of Google's rivals, was fined $1.5 billion（約2400億円） for copyright infringement related to its training materials, the largest payout in the history of U.S. copyright law. That settlement made approximately half a million writers eligible for payments of at least $3,000 each. However, many authors declined the Anthropic settlement to pursue independent legal action against other AI firms, suggesting the copyright conflict is far from resolved. Google did not immediately respond to a request for comment.

Google's Gemini lawsuit arrives amid a broader wave of copyright litigation against AI companies. While two California courts have already ruled that using copyrighted works for AI training constitutes "fair use" under U.S. copyright law, the legal landscape remains contested. Anthropic's $1.5 billion（約2400億円） settlement—the largest copyright payout in U.S. history—demonstrates that courts are willing to impose substantial penalties, even as the fair-use question remains unsettled. However, many authors have opted out of the Anthropic deal to pursue separate claims, signaling that a single precedent may not resolve the underlying conflict.

The Google case carries a distinctive angle that may complicate the fair-use defense. Unlike generic scraping of the internet, Google had an explicit relationship with publishers through Google Books, a program where authors and publishers provided works specifically for snippet-based search—a limited, licensable use. The plaintiffs' argument is that Google breached that trust by converting those same works into training data for a commercial AI product, all while an internal memo warned leadership that such use could be "highly problematic" and financially ruinous. Filed in New York federal court rather than California, the case will give a different judge the chance to interpret fair use in this specific context of pre-existing contractual relationships and internal knowledge of risk.