Google Knew the Risk: AI Copyright Lawsuit Cites Up to $100B in Fines

Major publishers and one of America’s best-known authors have taken Google to court over its AI ambitions, filing a class action lawsuit that accuses the company of raiding their copyrighted books to build its Gemini AI platform — without ever asking permission. The case lands in the U.S. District Court for the Southern District of New York, and it may prove far more complex than the copyright battles that came before it.

Key takeaways

• Hachette, Cengage, Elsevier, author Scott Turow, and S.C.R.I.B.E. filed a class action lawsuit against Google over alleged unauthorized use of copyrighted works to train Gemini.
• Google is accused of not only using books without permission but also removing or altering copyright information to conceal the practice.
• The lawsuit was filed in the U.S. District Court for the Southern District of New York, outside the California jurisdiction that has so far favored AI companies on fair use grounds.
• Plaintiffs previously gave Google access to their books solely for Google Books search snippets — not for AI training.
• An internal Google document allegedly warned that using copyrighted books for AI training could result in “$10Bs-$100Bs in potential fines.”

Publishers and Authors Sue Google Over AI Training Data

The plaintiffs are not a fringe group of aggrieved writers. Hachette, Cengage, and Elsevier are among the largest publishers in the world. Scott Turow is a celebrated novelist and longtime advocate for authors’ rights. Together with S.C.R.I.B.E., they represent a formidable coalition arguing that Google crossed a clear legal line when it fed their books into the data pipeline for Gemini.

The core allegation is straightforward: Google used copyrighted material to train its AI without authorization. But the lawsuit goes further, accusing Google of deliberately removing or altering copyright information on the works in question — a move the plaintiffs describe as an attempt to “conceal that its Gemini Models were trained on stolen materials.” That detail matters. It suggests not just careless use of protected content, but an intentional effort to obscure it.

A Trusted Relationship Turned Alleged Breach

What makes this case particularly pointed is the history behind it. Publishers and authors had a long-standing arrangement with Google: they provided access to their books so that Google Books could generate searchable snippets — short excerpts that help readers find titles without displaying full text. That was the deal. Users could see a few lines, not entire chapters.

The plaintiffs now allege that Google trained Gemini on copies of those very books, as well as on books uploaded to Google Play, without ever obtaining permission for that separate and far more expansive use. In their view, Google exploited a relationship built on a narrow, specific consent — and then quietly extended it to serve an entirely different commercial purpose.

That framing gives the lawsuit a dimension that purely adversarial AI copyright cases often lack. This is not just about an AI company scraping the open web. It involves a pre-existing contractual relationship, a defined scope of authorized use, and an alleged violation of that boundary.

The Fair Use Question and Why New York Changes the Calculus

The broader legal backdrop complicates the picture. Two early court decisions in California have ruled in favor of AI companies, finding that using copyrighted works for AI training constitutes fair use under U.S. copyright law — a statute that, notably, has not been updated since before the internet existed. Those rulings handed a provisional win to the AI industry and shaped expectations around how these disputes might be resolved.

But California is not the only courtroom in America. By filing in the U.S. District Court for the Southern District of New York, the plaintiffs are positioning their case before a judge who is not bound by those California precedents. The Southern District of New York carries its own authority and tradition in intellectual property matters, and it may weigh the fair use argument very differently — especially given the specific facts here, where the alleged misuse was not open-web scraping but the redirection of books shared under a clearly defined, limited-use agreement.

Anthropic’s $1.5 Billion Precedent

The industry already has one landmark data point. Anthropic was fined $1.5 billion for copyright infringement related to its AI training data — the largest payout in U.S. copyright history. Around half a million writers were eligible for payments of at least $3,000 from that settlement. Yet a significant number of authors rejected the money and opted out, choosing instead to pursue further legal action. Their reasoning: the settlement figure, however historic, may not reflect the true scale of the harm — or the precedent they want courts to establish.

That dynamic is worth watching. Authors who refuse settlements are not just making a financial calculation. They are signaling that they want courts, not companies, to define the rules of the road for AI training and intellectual property.

Google’s Internal Warning and Its Silence Now

Perhaps the most striking element of the lawsuit is a document the plaintiffs claim comes from inside Google itself. According to the filing, an internal Google document reportedly warned that using copyrighted books for AI training was “highly problematic for Google” and could result in “$10Bs-$100Bs in potential fines.”

If that document is authentic and enters evidence, it would suggest that Google’s own legal and risk teams had identified the exposure long before any lawsuit was filed. That is a qualitatively different posture than a company that simply miscalculated copyright risk. It raises harder questions about why, if the risk was internally understood, the practice continued.

Google did not respond to requests for comment on the lawsuit.

The silence is notable. With internal documents potentially in play and a New York court free to chart its own path on fair use, the Google Gemini lawsuit may force a reckoning that the California decisions — for all their AI-industry-friendly conclusions — did not fully resolve. The question of what publishers actually consented to when they partnered with Google on Books is now squarely before a federal judge, and the answer could reshape how every AI company negotiates access to copyrighted content going forward.

FAQ

Who are the plaintiffs suing Google in this case?

The plaintiffs are a group of major publishers and authors, including Hachette, Cengage, Elsevier, novelist Scott Turow, and S.C.R.I.B.E., who filed a class action lawsuit against Google over the alleged unauthorized use of their copyrighted works.

What is Google accused of in the lawsuit?

Google is accused of using copyrighted books without permission to train its AI platform Gemini, and of removing or altering copyright information on those works to conceal the alleged unauthorized use.

Where was the lawsuit against Google filed?

The lawsuit was filed in the U.S. District Court for the Southern District of New York, a jurisdiction outside California where earlier rulings have favored AI companies on fair use grounds.

What legal precedents exist regarding using copyrighted works for AI training?

Two court decisions in California have ruled that AI training with copyrighted works constitutes fair use under U.S. copyright law. However, those decisions do not bind courts in other jurisdictions, and the New York court handling the Google case may reach a different conclusion.

Article produced with the assistance of artificial intelligence and reviewed by the editorial team.