U.S. Publishers and Authors Sue Google Over AI Training with Books

Google is facing another copyright lawsuit from major U.S. publishers and bestselling authors regarding the training of its artificial intelligence models.
Three publishers—Hachette Book Group, Cengage Learning, and Elsevier—along with author Scott Turow, filed a class-action lawsuit for damages in the U.S. District Court for the Southern District of New York on July 14 (local time), alleging that Google used their copyrighted works without authorization to develop its AI model, Gemini.
Hachette is considered one of the five largest publishers in the United States, while Cengage Learning is an educational publisher known for titles such as "Principles of Economics" by N. Gregory Mankiw.
Elsevier is the publisher of the world-renowned scientific journal "Cell" and the medical journal "The Lancet."
In the complaint, the plaintiffs argued that Google improperly used millions of books and academic papers, which were provided for limited purposes such as "Google Books" and "Google Scholar," to train Gemini.
In other words, they claim that data provided under the condition that it only be used for search purposes was misappropriated for AI training.
They also criticized Google for acquiring training data through content accessible only after payment, as well as through so-called "pirate" websites.
The plaintiffs contend that Google intentionally removed author names and copyright notices to conceal the source of the data.
In particular, they pointed out that Gemini reproduced the table of contents and structure of "Principles of Economics" and provided verbatim responses from specific chapters. Regarding Scott Turow's novels, they noted that when prompted to "summarize the book so it can be read without buying it," the AI generated a detailed 2,000-word summary, including character profiles and the ending.
They argue that for AI training on copyrighted works to be recognized as "fair use," it must meet the requirement of "transformative use" that does not undermine the market value of the original work, which they claim Gemini failed to do.
Furthermore, the plaintiffs cited internal documents in the complaint, alleging that Google proceeded with using their works for AI training despite being aware that it could pose "fair use" risks.
These publishers and Scott Turow previously filed a similar copyright infringement lawsuit against Meta in May.
Google is also embroiled in similar lawsuits filed by authors, visual artists, and individual creators in 2023.
Beyond Google, OpenAI is currently engaged in legal battles with media outlets, including The New York Times, and the Authors Guild, while Anthropic is negotiating a settlement worth $1.5 billion (approximately 2.2 trillion won) regarding a lawsuit filed by authors.
(Photo: Provided by Google, Yonhap News)
※ Please note: This article was translated by AI and may contain errors.