Major publishers sue Google over Gemini AI training, claiming it used their books without permission

Key points

• Publishers including Hachette, Cengage, and Elsevier filed a class action lawsuit against Google in the U.S. District Court for the Southern District of New York.
• The plaintiffs allege Google stripped or altered copyright information on their works to hide that Gemini, its AI assistant, was trained on them without permission.
• An internal Google document cited in the lawsuit reportedly warned that using copyrighted books for AI training could result in "$10Bs-$100Bs in potential fines."
• Anthropic, a rival AI company, was fined $1.5 billion in a separate copyright case, the largest copyright payout in U.S. history.
• Two earlier California courts ruled that training AI on copyrighted material counts as "fair use" under current U.S. copyright law, though that law has not been updated since before the internet existed.

A group of major publishers and authors has sued Google, claiming the company used their copyrighted books to train Gemini, its AI assistant (the chatbot that competes with ChatGPT). The lawsuit was filed in federal court in New York.

The plaintiffs include publishing giants Hachette and Elsevier, textbook publisher Cengage, thriller author Scott Turow, and a writer advocacy group called S.C.R.I.B.E. They say Google did not just copy their books. They also claim Google deliberately removed or altered copyright notices to hide the fact that it had done so.

The complaint has a specific backstory. Many of these publishers had already shared their books with Google for a narrow purpose: powering Google Books, the search tool that lets you find a title and read a short excerpt, not the full text. The lawsuit argues Google then quietly used those same files, plus books uploaded to Google Play, to train Gemini. No permission was asked. No payment was made.

And here is the detail that will make any lawyer wince. The lawsuit cites an internal Google document that apparently warned the company that training on copyrighted books could be "highly problematic for Google" and might cost the company tens to hundreds of billions of dollars in fines. Google proceeded anyway, the plaintiffs say.

Google has not commented publicly on the case.

Does this lawsuit have a realistic chance of winning?

It is genuinely unclear, and that uncertainty matters for every writer and reader. Two California courts have already sided with AI companies, ruling that using copyrighted work for AI training counts as "fair use," the legal doctrine that allows limited use of someone's work without payment (think a teacher photocopying a poem for class). But those rulings are not automatically binding on a New York judge, and the publishers' relationship with Google here is more complicated than most cases. They gave Google their books for one specific job, and they say Google used them for something else entirely.

Separately, AI company Anthropic settled a copyright lawsuit in 2025 for $1.5 billion, the largest copyright settlement in U.S. history, first reported by TechCrunch AI. Around 500,000 writers qualified for payments of at least $3,000 each, though many opted out to keep their right to sue independently.

For working authors and publishers, the practical takeaway is simple. The legal question of who owns the words that feed AI systems is still wide open. Courts are reaching different conclusions, and a New York ruling in this case could shift the balance. Watch this one.