A group of major publishers and authors has filed a class action lawsuit against Google, claiming the company used their copyrighted books to train its AI model, Gemini, without permission. The plaintiffs include Hachette, Cengage, Elsevier, novelist Scott Turow, and the author advocacy group S.C.R.I.B.E. The lawsuit was filed in the U.S. District Court for the Southern District of New York.
The allegations go beyond unauthorized use. As reported by TechCrunch, the plaintiffs also claim Google deliberately removed or altered copyright information on the works to hide the fact that its Gemini models were trained on stolen material. That is a serious accusation, and it moves the case well beyond a standard copyright dispute.
Google did not immediately respond to a request for comment.
This lawsuit fits into a much larger pattern. Publishers, authors, musicians, and visual artists have been filing copyright claims against AI companies for years, targeting Google, Meta, OpenAI, and Anthropic among others. The core question in all these cases is the same: does training an AI on copyrighted material count as “fair use” under U.S. copyright law?
So far, the courts have not given a clear answer. Two early decisions in California sided with the AI companies, ruling that using copyrighted works for training does qualify as fair use. That law, it is worth noting, has not been updated since before the internet existed. Those rulings are not good news for plaintiffs, but they are also not the final word. Because this case is filed in New York, a different judge will weigh in, and the legal details here are specific enough to produce a different outcome.
The Google case has a wrinkle that other AI training lawsuits do not. Publishers and authors have a long-standing relationship with Google through Google Books, a program where they provided copyrighted works specifically so Google could make them searchable. The search results only show short snippets, not full books. The plaintiffs argue that Google then took those same books, along with titles uploaded to the Google Play store, and used them to train Gemini, without ever getting permission to do so.
“Google illegally copied works from all these scope-limited programs for AI training, knowing it lacked authorization to do so,” the lawsuit states.
The plaintiffs also point to an internal Google document that allegedly acknowledges the risk. According to the lawsuit, the document warns that using copyrighted books for AI training could be “highly problematic for Google” and could lead to fines in the range of “$10Bs-$100Bs.”
That document, if it holds up, could be significant. It suggests Google was aware of the legal exposure and proceeded anyway. Courts tend to look less favorably on defendants who knowingly took a risk at someone else’s expense.
The broader stakes here are real. The Anthropic case offers a reference point: the company was fined $1.5 billion for using pirated works in its training data, the largest copyright payout in U.S. history. Around half a million writers qualified for payments of at least $3,000, though many opted out to preserve their right to sue separately. That case shows these lawsuits can result in meaningful consequences, even when the legal path is uncertain.
What happens in New York could matter a great deal, not just for Google, but for how the entire AI industry handles training data going forward.
