Three Publishers Sue Google Over Copyright Infringement in Gemini Training

Three major U.S. publishers and an author have initiated a class-action lawsuit against Google. According to a report by Anna Washenko of Engadget, Hachette Book Group, Cengage Learning, Elsevier, and author Scott Turow have filed a lawsuit in U.S. federal court, alleging that Google used copyrighted works without authorization as training data for its AI system “Gemini.”

Lawsuits concerning copyright infringement by AI companies over training data are becoming increasingly common. The publishers allege that “Google has replicated millions of copyrighted works without permission and without paying any compensation to authors or publishers.” This lawsuit highlights the legal challenges arising from the disconnect between the AI industry and existing copyright laws.

Details of the Lawsuit

According to the complaint, the plaintiffs argue that Google’s AI training practices violate copyright laws. A particularly noteworthy claim involves the removal of copyright management information (CMI). The plaintiffs allege that “Google stripped CMI from the works it copied, making it easier to conceal the sources of its training data and to use them without obtaining licenses.”

CMI includes details such as a work’s title, author, copyright holder, and usage terms. Under the Digital Millennium Copyright Act (DMCA), the removal of CMI is prohibited. If this claim is upheld, it could lead to additional legal liabilities beyond straightforward copyright infringement.

The plaintiffs also assert that Gemini facilitates copyright infringement. They claim that Gemini generates derivative works on behalf of users, potentially replacing the original copyrighted material, and criticize Google for failing to implement effective safeguards against this issue.

Past Lawsuits and Challenges

The publishing industry has consistently taken legal action against AI companies over the use of training data. Similar groups have already filed class-action lawsuits against Meta. However, lawsuits based on copyright infringement have rarely been successful.

In a notable case from 2025, another group of authors reached a $1.5 billion settlement with Anthropic over copyright infringement, but the presiding judge rejected the settlement, calling it “far from sufficient.” Similarly, a lawsuit against Meta was dismissed last year. Another group of authors also sued Apple for unauthorized use of their works in AI training.

These cases highlight the unresolved legal relationship between AI training and copyright law. Courts continue to grapple with the fine line between “learning” and “copying,” and clear legal standards have yet to emerge.

Within the AI industry, the primary point of contention is whether using copyrighted works as training data qualifies as “fair use.” U.S. fair use laws assess factors such as transformative use and market impact, making the determination of fair use for AI training a case-by-case issue.

Publishers’ Claims

The publishers emphasize that Google’s actions are deliberate. The lawsuit states, “Google was fully aware that its actions violated copyright laws and replicated millions of copyrighted works without permission.” Since the controversy surrounding Gemini’s training dataset in 2024, Google has faced multiple copyright lawsuits.

A key point of contention is the possibility that Google collected some of its training data from its search engine index without notifying or entering into licensing agreements with publishers or authors.

The publishing industry has long called for transparency regarding how their works are used in AI training. However, AI companies have consistently refused to disclose the composition of their training datasets, citing trade secrets, which has fueled distrust across the industry.

Editorial Opinion

The challenges posed by copyright lawsuits over AI training underscore the difficulty of adapting legal frameworks to rapid technological innovation. While the success of this lawsuit remains uncertain, the publishers’ arguments are consistent, and their claim regarding the removal of CMI could be particularly strong from a legal standpoint. In the short term, this lawsuit may inspire similar actions by other rights holders. Depending on how Google responds, it could lead to greater transparency in training data or expedite licensing agreements.

In the long term, new licensing models between AI companies and copyright holders will need to be established. These could include models where publishers directly charge for usage, licenses mediated by industry organizations, or regulatory frameworks mandating transparency in training data. The AI training process remains a technical “black box,” complicating legal judgments. The extent to which courts can accurately understand and rule on the technical realities will be a key factor shaping the future of the AI industry.

References

• “Three publishers challenge Google over AI copyright infringement”, by [email protected] (Anna Washenko) — Engadget, 2026-07-14T20:28:41.000Z (ARR)
• Source URL: https://www.engadget.com/2215206/three-publishers-challenge-google-over-ai-copyright-infringement/