The reckoning between the book world and the makers of artificial intelligence has reached Google. A coalition of publishers, joined by the novelist Scott Turow, has sued the company over what it calls willful copyright infringement in the training of Google's Gemini AI models.
The allegations
The lawsuit, brought by publishers including Hachette Book Group, Cengage and Elsevier, alleges that Google copied vast quantities of copyrighted books to train its AI without a license or payment. The plaintiffs contend that Google drew on works originally shared with it for the limited purposes of Google Books, as well as material scraped broadly from the web, including, they say, from pirate sources and behind paywalls. They also allege Google removed or altered copyright information to obscure what its models had been trained on. The suit points to internal company documents that, according to TechCrunch's account, flagged the use of copyrighted books for training as legally risky.
The publishers are seeking damages and a court order to stop the alleged infringement, and want copies made for training destroyed.
Google's likely defense
Google has not addressed the specific claims but is widely expected to argue "fair use," the copyright doctrine that permits limited use of protected works without permission. AI companies contend that training a model on text, like a person learning from reading, is transformative and legal. That argument has had some success: courts have ruled in favor of AI developers in a couple of recent training cases, as TechCrunch noted.
The publishers' case has a wrinkle they hope will distinguish it. Because many of them had an existing relationship with Google through Google Books, providing works for a specific, bounded purpose, they argue Google exceeded what was agreed and cannot claim the same latitude as a company scraping the open web.
A widening front
The Google suit is the latest in a fast-multiplying set of legal challenges over how AI systems are built. OpenAI, Meta and Anthropic have all faced copyright claims from writers and publishers, with mixed results, including one settlement that ranked among the largest in copyright history. At issue across all of them is a question the law has not settled: whether feeding the world's books, articles and art into a machine to build a commercial product is a fair use or a mass infringement.
For authors and publishers, the fear is existential, that AI trained on their work will ultimately compete with it, without their consent or a cut of the proceeds. For Google and its rivals, the stakes are the raw material of the AI boom itself. The Google case, like the others, could help decide whether the companies must pay to license what they read, and it is unlikely to be the last word.
