Three prominent publishers, alongside acclaimed novelist Scott Turow and his company S.C.R.I.B.E., have initiated a proposed class-action lawsuit against Google. The core of their accusation is that Google allegedly utilized millions of copyrighted books and journal articles, originally provided for services like Google Books, Play Books, and Scholar, to train its advanced artificial intelligence model, Gemini, without obtaining the necessary permissions. As of the latest reports, no court has rendered a definitive ruling on these contentious claims, leaving the legal and ethical implications of AI training data at the forefront of a burgeoning debate.
The lawsuit, filed on July 10th in the U.S. District Court for the Southern District of New York, brings together Hachette Book Group, Cengage Learning, and Elsevier, renowned entities in the publishing world, with Scott Turow and S.C.R.I.B.E., representing authors and intellectual property holders. The Association of American Publishers (AAP) publicly announced the filing on the same day, amplifying the industry’s collective concern. At the heart of their legal argument lies the assertion that the literary works supplied to Google’s platforms were intended for specific, limited purposes—such as search previews or user access within designated services—and that their repurposing for the development of a commercial AI model constitutes a significant violation of copyright. The plaintiffs further allege that Google’s data acquisition practices extended to unauthorized copying of works obtained through web scraping, including content harvested from pirated websites and content behind paywalls of subscription libraries. Google has maintained a policy of not commenting on ongoing litigation, and consequently, has not offered a public statement regarding this specific complaint.
The Allegations Detailed in the Complaint
The formal complaint outlines four distinct counts against Google. The first three counts center on allegations of unauthorized reproduction in violation of the Copyright Act. These specifically address:
- Google Books and Related Services: The alleged unauthorized copying of works provided through Google Books and other Google-facilitated services.
- Web Scraping Downloads: The claim that Google copied copyrighted materials acquired through broad web scraping, irrespective of their original hosting or licensing.
- Copying During Training: The accusation that the process of training Gemini itself involved the unlawful reproduction of these literary works.
The fourth count accuses Google of violating the Digital Millennium Copyright Act (DMCA) by allegedly removing or altering copyright management information associated with the works.
The plaintiffs are seeking substantial remedies, including monetary damages, an injunction to prevent further alleged infringement, a comprehensive accounting of all works used by Gemini for its training, and court orders mandating the deletion of any unauthorized copies. Notably, the lawsuit cites what it describes as internal Google documents, offering a glimpse into the company’s internal deliberations. One such internal communication reportedly labeled the use of books from Google Play Books for AI training as "highly problematic for Google," with potential financial repercussions estimated in the "tens of billions to hundreds of billions of dollars." Another quote attributed to Gemini’s lead engineer allegedly stated, "we don’t do deals for data we already have or already possess." It is crucial to note that these documents have not been made public, and these quotes are presented as part of the plaintiffs’ legal filing.
Navigating the Nuances of Data Sourcing and Crawler Controls
A significant point of contention in this legal battle revolves around how Google obtained the data used to train Gemini and whether existing controls were bypassed or rendered irrelevant. The lawsuit touches upon the use of Google-Extended, a robots.txt token designed to restrict content crawling for specific Google services, including certain AI training and grounding uses. However, the methods of data acquisition alleged in the complaint appear to circumvent this mechanism.
The publishers and authors contend that the books were supplied directly to Google through contractual agreements. In such cases, a website’s robots.txt file, which governs how search engine crawlers interact with publicly accessible web content, would not be applicable. These agreements, the plaintiffs argue, defined the permitted uses of the content, and AI training was not among them.
Furthermore, the claims related to web scraping highlight a different pathway for data acquisition. The complaint asserts that copies of copyrighted works appeared in datasets like Common Crawl after being hosted on unauthorized platforms, including pirate sites and subscription-based digital libraries. Because these alleged copies were hosted on domains distinct from the original rights holders’ websites, a robots.txt file on the original site would not govern their subsequent aggregation and use by Google through web scraping.
This distinction is critical, as it challenges Google’s potential reliance on defenses like "fair use," which often considers the nature of the use and the source of the data. Google, in a policy paper released on June 25th, argued that training AI models on publicly available web data constitutes a "transformative, non-expressive use" protected by fair use principles. The paper also acknowledged the existence of machine-readable controls, such as Google-Extended, that allow website owners to opt out of such data usage. However, the plaintiffs in this case allege that the data in question was acquired through channels that fall outside the scope of these controls, raising questions about the applicability of Google’s stated policies and defenses.
Industry Reactions and Broader Implications
The lawsuit has sent ripples throughout the publishing and technology industries, highlighting a growing tension between content creators and AI developers. The Association of American Publishers, in its announcement, emphasized the "willful copyright infringement" by Google in its pursuit of developing advanced AI models. This sentiment is echoed by many in the creative sector who fear that their intellectual property is being exploited without compensation or consent, potentially undermining the economic viability of their work.
The issue of fair use is central to this legal dispute. While Google argues that its use of copyrighted material for AI training is transformative and falls under fair use protections, copyright holders maintain that such use constitutes infringement, especially when the original intent of providing the material was for different purposes. The question of whether permission granted for one type of digital access implicitly allows for AI training is a novel and complex legal challenge.
The complaint’s inclusion of internal Google documents, if proven authentic and admissible, could provide significant leverage for the plaintiffs. These internal communications might reveal a level of awareness within Google regarding the potential legal and financial risks associated with using copyrighted material for AI training without explicit agreements.
A Developing Timeline of AI and Copyright Disputes
This lawsuit against Google is not an isolated incident. It emerges within a broader context of legal challenges and industry-wide discussions concerning the use of copyrighted content for AI training.
- Early 2023: Reports began to surface of authors and artists discovering their works being used to train large language models without their consent. This sparked initial outrage and calls for greater transparency and compensation.
- Mid-to-Late 2023: Several class-action lawsuits were filed against AI companies, including OpenAI and Meta, by authors and artists alleging copyright infringement. These lawsuits raised similar questions about the legality of using vast datasets scraped from the internet.
- June 25, 2024: Google releases a policy paper arguing for the legality of AI training on public web data under fair use.
- July 10, 2024: The publishers and Scott Turow file their class-action lawsuit against Google in the U.S. District Court for the Southern District of New York.
- Prior Rulings: In June 2025, two Northern California district courts issued rulings that touched upon AI training and fair use. One ruling denied summary judgment on claims involving pirated library copies used for AI training, while another stressed that its decision was specific to the plaintiffs and the evidence presented. These rulings offer a mixed and evolving legal landscape.
The plaintiffs in the Google case stated they chose to file in New York after initially considering intervening in the ongoing In re Google Generative AI Copyright Litigation in California. They believe their new suit preserves claims that fall outside the scope of that proposed class, suggesting a strategic approach to navigating the complex legal terrain.
The Significance of Crawler Controls and Data Provenance
The discussion around robots.txt and Google-Extended is particularly relevant. While 79% of top news sites reportedly block at least one AI training bot, as indicated by BuzzStream data in January, this statistic primarily pertains to content directly accessible via web crawling. The core of the current dispute lies in data acquired through different means: explicit agreements for services like Google Books and Play Books, and extensive web scraping that may have captured content from less conventional sources.
This lawsuit underscores a critical distinction: contractual agreements for specific digital services may not automatically extend to broader AI training purposes. Similarly, data aggregated from the open web, even if publicly accessible, can still be subject to copyright protections. The provenance and method of data acquisition are likely to be pivotal in determining the legal outcomes.
Looking Ahead: Legal Precedents and Future Implications
The legal battle ahead is complex, with significant implications for the future of AI development and the rights of content creators. The core question—whether permission for one use of a copyrighted work implicitly grants permission for its use in training AI models—remains largely unanswered by established legal precedent.
The outcome of this lawsuit, and others like it, could shape how AI models are trained and how intellectual property is protected in the digital age. It could lead to new licensing frameworks, greater transparency in data sourcing, and potentially, significant financial repercussions for companies that do not adequately address copyright concerns.
The next procedural step in this case will be Google’s response to the complaint, which could involve filing an answer to the allegations or a motion to dismiss the lawsuit. The ensuing legal proceedings will be closely watched by publishers, authors, technology companies, and legal experts alike, as they navigate the uncharted territory of artificial intelligence and intellectual property law. The ultimate resolution will likely set important precedents for how creative works can be utilized in the development of next-generation technologies.
