I woke to an alert about a book that sounded like my voice but had never been written. You feel that sudden wrongness—the familiar phrasing, the absent byline. That moment is what this lawsuit turns into a legal fight.
I’ve tracked publishing wars and copyright fights long enough to tell you this is not a sideshow. Hachette Book Group, Cengage Learning and Elsevier filed suit in the U.S. District Court for the Southern District of New York, saying Google used their copyrighted books and journal articles to train its AI, Gemini. Author Scott Turow has joined the complaint and the plaintiffs are seeking class-action status.
Did Google illegally copy books to train Gemini?
The complaint says yes. It accuses Google of reproducing “millions of copyrighted works without permission, without providing any compensation to authors or publishers, and with full knowledge that its conduct violated copyright law.” The plaintiffs point to text scraped from legitimate sources and from “known pirate sources” as the raw material that taught Gemini to write and summarize.
The view from a bookstore counter: a reader asks how a best-selling novel appears in an instant
On the shop floor you can see it: someone reading a machine-generated summary beside a shelf of printed books. The suit argues Gemini doesn’t just summarize — it can output verbatim or near-verbatim passages, replacement chapters for textbooks, and alternative versions of novels. The complaint says those outputs compete directly with the original works.
Google’s chatbot, the suit claims, can produce a 100-page murder mystery in about 20 minutes for roughly $0.39 (€0.36). That speed and scale, the publishers say, undercuts the market for human-authored books. Gemini reportedly has more than 650 million monthly active users, and Google’s parent company generated approximately $100 billion (€93 billion) in revenue in October 2025 — figures the plaintiffs use to argue Google could have licensed material rather than copying it.
What do publishers want from the lawsuit against Google?
They want payment, control, and an industry reset. Hachette, Elsevier, Cengage and Turow say the infringement was willful and that damages and an injunction are necessary to stop further harm. The suit points out that other publishers have negotiated licensing deals — for example, HarperCollins signed a deal in 2024 with Microsoft to supply books for AI training — and argues Google could have done the same.
The result is an AI system that competes directly with Plaintiffs’ and the Class’s works in the market. Those substitutes take multiple forms, including verbatim and near-verbatim copies of portions or entire works, replacement chapters of academic textbooks, summaries and alternative versions of famous novels, and inferior knockoffs that copy creative elements of original works. Gemini even tailors outputs to mimic the expressive elements and creative choices of specific authors.
A professor in a lecture hall: students hand in essays that read suspiciously polished
When academic publishers notice content from journals like The Lancet or Cell reappearing in AI outputs, the alarm shifts from commerce to the integrity of scholarship. The complaint accuses Google of copying journal articles that are foundational to research and teaching, then using them to power a model that can produce substitute textbook chapters.
Elsevier and Cengage, both heavyweights in academic and educational publishing, joined Hachette in saying that automated replicas threaten both revenue and the incentive to create new works. The plaintiffs put it plainly: copyright law applies to AI companies the same way it applies to any other company that uses protected work.
How does this affect authors like Scott Turow?
For established authors such as Turow, the harm is both financial and reputational. The suit alleges Gemini can mimic expressive choices of specific authors, producing outputs that borrow voice, structure and creative elements. That reduces the unique value of an author’s style and complicates how readers and buyers decide what to buy.
Turow and the publishers previously sued Meta on similar grounds earlier this year, so this case fits into a broader industry pushback against large tech firms that train models on scraped text. The plaintiffs want to stop unlicensed copying, recover damages and force a market where AI companies pay for training data or risk court-ordered limits.
I’ll be blunt: the legal theory is straightforward to the plaintiffs — copying without a license is unlawful — but the stakes are large. If courts side with publishers, we could see licensing markets and fees for training data. If courts side with the tech firms, AI output will be harder to tie to traditional copyright, and authors will face a new competitive landscape.
Google has not responded to emailed questions about the lawsuit; Gizmodo has said it will update coverage if Google replies. Meanwhile, the complaint highlights that Google’s AI business is a major revenue driver and argues the company could have paid for the content it used.
One last note: think of Gemini as a photocopier gone sentient—accurate, fast, and indifferent to bylines—and the publishing industry as a cathedral whose stained-glass is being copied into cheap panes. The legal fight will decide whether the copies are allowed to stand.
What should readers and creators expect next: a settlement, a landmark ruling, or years of appeals that reshape how AI models are trained?