The Google Gemini copyright lawsuit filed in the U.S. District Court for the Southern District of New York arrives with an unusual twist: the publishers suing Google spent years actively cooperating with the company before deciding it had betrayed them.
Plaintiffs Hachette, Cengage, Elsevier, author Scott Turow, and the organisation S.C.R.I.B.E. allege that Google used copyrighted works to train its Gemini AI models without authorisation, and that it deliberately removed or altered copyright information to conceal the fact. The Guardian reports the lawsuit characterises this as ‘one of the most prolific infringements of copyrighted materials in history.’
According to Al Jazeera, the suit alleges that ‘at no point did Google inform authors and publishers that Google was copying their works as source material to develop and train AI models.’ That silence is the crux of the case: the publishers are not strangers who got scraped. They had a relationship.
A Decade of Book Deals, Then a Betrayal Claim
Publishers and authors have a long history with Google Books, providing copyrighted works specifically so the company could make titles searchable. Users see short snippets and bibliographic details, not whole books. The plaintiffs allege Google then used those same copies, plus books from Google Play, to train Gemini without ever asking.
Publishing Perspectives reports the allegations extend further, to works scraped from behind paywalls, from what the lawsuit calls ‘known pirate sources,’ and from Google Scholar. The lawsuit is also explicit that the 2015 U.S. appeals court ruling affirming Google’s book-scanning project does not, by its terms, authorise or ‘deem it fair use’ for Google to copy those same works for the separate purpose of training commercial AI models.
The plaintiffs also cite an internal Google document allegedly stating that using copyrighted books for AI training could be ‘highly problematic for Google’ and might result in ‘$10Bs–$100Bs in potential fines.’ Google had not responded to a request for comment at the time of publication.
The Google Gemini Copyright Lawsuit in a Crowded Legal Field
This action is the eighth copyright lawsuit filed against Google relating to AI. Two earlier suits have been consolidated as In re Google Generative AI Litigation, and the International Publishers Association confirms the publishers originally intended to join that consolidated case as intervenors, then filed separately to preserve claims falling outside the scope of the existing class.
It sits within a broader wave of litigation against AI companies including Meta, OpenAI, and Anthropic. Most cases remain unresolved, and two early California decisions have favoured the AI companies, ruling that training on copyrighted works constitutes fair use under copyright law that predates the internet. But fair use is not a blanket shield.
In June 2025, U.S. District Judge William Alsup ruled on summary judgment in the Anthropic case that training on legally acquired books was fair use, but denied Anthropic’s fair-use argument where piracy was involved, according to the Authors Guild. The distinction mattered enormously to the outcome.
The Wolters Kluwer copyright blog reports that court documents showed Anthropic downloaded over 7 million books from the LibGen and PiLiMi pirate sites for AI training purposes. Anthropic then reached a $1.5 billion settlement in August 2025 in the case captioned Bartz v. Anthropic PBC, the largest payout in U.S. copyright history. The settlement covers rightsholders of books obtained from those pirate sources that carried ISBN or ASIN numbers and were registered with the U.S. Copyright Office. After administration fees and legal costs, with roughly a quarter of the total going to lawyers, the remainder is split between authors and publishers.
Around half a million writers were eligible for payments of at least $3,000. A number of authors declined, preferring to preserve their right to pursue further legal action, according to Lieff Cabraser, the plaintiffs’ firm in that case.
The California fair-use rulings are not an automatic precedent. The new Google case lands in the Southern District of New York, before a different judge, with a fact pattern specifically engineered around the betrayal of an existing, consensual relationship. The California outcomes tell us the legal theory is contested. They don’t tell us how a New York judge weighs a case where the publishers can point to years of specific agreements about what their books were being used for.
Watch whether Google files a fair-use motion early, and whether the court allows the piracy and the scope-limitation claims to proceed to discovery. If they do, the internal document flagging potential fines of $10 billion to $100 billion becomes a live exhibit, not a footnote.
