Publishers and author sue Google over Gemini AI training
Hachette, Cengage, Elsevier and Scott Turow allege Google used copyrighted books without permission to train Gemini.
By Sofia Marchetti · World Affairs Correspondent
4 min read
Three major publishers and author Scott Turow have sued Google in federal court in New York, accusing the company of using copyrighted books and other works without permission to build its Gemini artificial intelligence models. The case adds Google to a growing list of AI companies facing claims from writers, publishers and media groups over how training data is obtained and used.
Hachette Book Group, Cengage Learning, Elsevier and Turow filed the complaint on Friday, according to Al Jazeera. The nearly 60-page lawsuit alleges Google made deliberate decisions to avoid copyright protections while developing Gemini.
The complaint says Google first obtained books through Google Books and other services for limited purposes, then used those materials as training data beyond the scope of those arrangements. It also alleges Google downloaded large-scale web data that included material from pirate sites and content behind legitimate paywalls.
The plaintiffs say Google copied works without authorization and continues to do so. They also allege the company knew the legal exposure, citing internal documents that described the use of books to train AI systems as highly problematic and warned of possible fines as high as $100bn.
According to the complaint, Google did not tell authors and publishers that their works were being copied as material for AI development and training. Google did not respond to Al Jazeera’s request for comment.
Fair use fight likely central
Kirk Sigmon, a founding partner at KellDann Law who works on technology and intellectual property issues, told Al Jazeera that the case could test whether Google can rely on fair use if the plaintiffs show the books were obtained unlawfully. He said proving what was or was not included in an AI training set can be difficult.
The lawsuit follows an earlier effort by Hachette and Cengage in February to join an existing class action first brought by authors in 2023. Hachette said in a statement that the new complaint shows authors and publishers are aligned in seeking to protect works across fiction, nonfiction, children’s books, memoirs, poetry, educational material and scholarly articles.
Michael Goodyear, an associate professor at New York Law School, told Al Jazeera that many of the copyright cases against AI companies follow a similar pattern: plaintiffs allege copyrighted works were copied and used for training, and some also argue that AI outputs infringe protected material.
Oli Huggins, CEO of ExpertEdge and vice president of partnerships at Packt Publishing, told Al Jazeera that tracing copyrighted material inside an AI model is a major challenge. He said a model may appear familiar with a book without reproducing enough exact text to prove which copy was used or how it affected the system.
Huggins also said current licensing offers from AI companies are unattractive for publishers, citing offers that can value a permanent AI training license at about $10 per title.
Broader legal pressure on AI firms
The Google case lands amid wider litigation over AI training in publishing, news and music. Authors including George R.R. Martin and the Authors Guild are pursuing a case against OpenAI, and Al Jazeera reported that a federal judge in October rejected OpenAI’s attempt to dismiss it.
A separate case against Meta went the other way. In 2025, a federal judge ruled that AI training in a lawsuit brought by authors led by Richard Kadrey met the legal standard for fair use, according to Al Jazeera.
News organizations have also sued AI companies. CNN filed a complaint against Perplexity in May alleging unlawful copying of more than 17,000 stories, while 17 news organizations, including The New York Times, accused OpenAI last week of withholding evidence in a copyright dispute first brought by the Times in 2023.
In music, Hagens Berman has filed a class action against AI music generator Suno, alleging use of independent musicians’ work without consent. Universal Music Group sued Anthropic in January, alleging the company used 20,000 songs to train Claude without permission.
Goodyear told Al Jazeera that courts still have not fully resolved who is liable when an AI system produces material that appears copied, particularly if a user is trying to induce infringement.
This story draws on original reporting from Al Jazeera.