Hachette Book Group, Cengage Learning, Elsevier, and author Scott Turow have filed a lawsuit in federal court in New York against Google, accusing the Silicon Valley tech giant of copyright infringement while training its Gemini AI model.
“Through a series of deliberate choices in developing Gemini1, Google intentionally circumvented this long-standing system designed to protect copyright and compensate authors and publishers,” the nearly 60-page complaint filed Friday says.
Recommended stories
list of 4 itemsend of list
The lawsuit claims that Google first copied the books as source material through Google Books, and that the company used “books obtained for strictly limited purposes in connection with Google Books and other Google services.” It also claims that Google has “downloaded web scrapings of virtually the entire Internet, including from known pirated sources and from behind legitimate paywalls.”
Further, the lawsuit alleges that Google has been copying those works without permission to train its AI models and continues to do so even though their use is said to be outside the scope of existing agreements.
The lawsuit alleges that the company was well aware of the legal risks and was warned in internal documents that using the book to train AI models was “highly problematic for Google” and could lead to fines of up to $100 billion.
“At no point did we notify authors and publishers that Google was copying their work as source material for the development and training of AI models,” the complaint alleges.
“In short, any fair use [a legal doctrine that allows limited use of copyrighted material without permission for purposes like reporting, education, and research in a limited scope] “Gemini’s claims are probably vindicated by the fact that they obtained the books illegally,” Kirk Sigmon, founding partner in technology and intellectual property law at Keldon Law Firm, told Al Jazeera. “This is an interesting issue with many complex aspects, in no small part because it is difficult to prove what was or was not included in the training corpus.”
The lawsuit follows a previous attempt in February by Hachette Book Group and Cengage to join an existing class action lawsuit originally filed in 2023 by a group of authors.
“The scope of the complaint highlights that authors and publishers are united in the goal of protecting valuable intellectual property rights in fiction, nonfiction, children’s books, memoirs, poetry, and educational and scholarly articles across thousands of subject areas and research and developments,” Hachette said in a statement after the complaint.
Google did not respond to Al Jazeera’s request for comment.
wave of lawsuits
This is not the first lawsuit authors and book publishers have filed against AI companies for alleged copyright infringement.
Also pending is a lawsuit against OpenAI brought by writers including Game of Thrones author George R.R. Martin and the Writers Guild. In October, a federal judge rejected OpenAI’s attempt to dismiss the lawsuit.
However, another lawsuit brought against Meta by a group of authors was not in their favor. In 2025, a group of authors led by Richard Kadry claimed that the parent companies of Facebook and Instagram used copyrighted books to train AI models. A federal judge has ruled that AI training meets the legal requirements for “fair use.”
“They are [the copyright lawsuits] Usually everything follows the same roadmap,” Michael Goodyear, an associate professor at New York Law School, told Al Jazeera.
“The basic argument is that you took a copyrighted work and used it for training. These are illegal copies, and that’s the training argument. Some argue more specifically about copyright infringement.”
However, Packt, the CEO of ExpertEdge and the publisher whose title is used in the Humanity copyright infringement case, Oli Huggins, vice president of partnerships at Publishing, told Al Jazeera that the company has been approached by AI companies to pay to train using its data, and said the challenge is that once the information is used for training, it is difficult to prove whether the output constitutes copyright infringement.
“Proving what happened inside the model is another fundamental difficulty. Once the egg is baked into the cake, it’s hard to identify it, quantify its contribution, or prove exactly which copy of the book was used. “It is very difficult for a model to demonstrate familiarity with the work without reproducing enough verbatim text to establish the evidence the plaintiff needs,” Huggins told Al Jazeera.
Mr Huggins said the current licensing arrangements were not sustainable for publishers.
“The economic situation remains very unattractive. With the offers currently in circulation, the value of a perpetual license for AI training is around $10 per title.”
Across content-driven industries such as news and music, lawsuits against AI companies for alleged copyright infringement have sprung up.
According to a complaint filed in May, CNN filed a lawsuit against Perplexity, alleging that the company illegally copied more than 17,000 stories to train its models, resulting in content that was “identical or substantially similar to CNN’s content.”
Last week, 17 news organizations, including the New York Times, accused OpenAI of withholding evidence from a 2023 lawsuit filed by the AI company led by Sam Altman for copyright infringement during ChatGPT training.
In the music field, Hagens Berman, a prominent class action law firm, filed a class action lawsuit alleging that AI music generation company Suno trained its models on the work of independent musicians without their consent.
Universal Music Group sued Anthropic in January, accusing Dario Amodei’s company of copyright infringement by using 20,000 songs in Claude Modell’s training without permission.
But Goodyear says the question of who is ultimately responsible for the copies that appear to have been copied remains, and the courts have not yet answered it.
“If a user is actively trying to compromise a model, it could ultimately mean that the user is in trouble, not the AI system,” Goodyear said. “This is still an open question and has not been seriously addressed by U.S. courts.”
