Introduction
The legal proceedings involving Anthropic PBC and the class certification in the case of Bartz v. Anthropic highlight significant issues surrounding copyright use in AI training. The case addresses the use of copyrighted materials, both lawfully acquired and pirated, in training AI models, with potential implications for the AI industry and copyright law.
Description
Anthropic PBC has filed a Rule 23(f) petition for an interlocutory appeal regarding class certification in the case of Bartz v [1]. Anthropic [1] [2] [3] [4] [5] [6] [7], seeking to pause the copyright lawsuit until the Ninth Circuit addresses its appeal. This petition follows a request to District Judge William Alsup for permission to appeal a prior ruling that certified a class of authors whose works were allegedly used without permission in training Anthropic’s Claude model, which includes approximately seven million pirated books [3]. The court’s recent ruling on class certification determined that it is appropriate for certain groups of plaintiffs, specifically those who are beneficial or legal copyright owners of books downloaded by Anthropic from sources like Library Genesis and PiLiMi [2]. However, certification was denied for plaintiffs associated with Book3 and scanned books [2]. Judge Alsup’s decision on class certification and the fair use issue is expected to significantly influence the legal landscape surrounding AI systems [7].
In a notable ruling [2], the court found that Anthropic’s use of lawfully acquired copyrighted books to train its Claude chatbot constituted fair use [5], emphasizing the transformative nature of this usage [5]. The ruling clarified that while the long-term storage of pirated books is infringing and not protected under fair use, the transformative purpose of the training, the nature of the works involved, and the necessity of using the full text for training were key factors, despite the output not being verbatim. This distinction between lawfully acquired and pirated materials will be a critical aspect of the upcoming trial.
The trial is scheduled to commence on December 1, 2025 [1], with a key focus on the potential statutory damages that a jury in San Francisco may award to the authors of the Bartz book for the unauthorized downloading of their works by Anthropic [1]. The emergence of a class action lawsuit significantly increases the stakes for AI companies, as statutory damages could lead to claims in the hundreds of millions or even billions [3], potentially jeopardizing the viability of many firms in the sector [3]. A certified class could increase pressure on Anthropic to reach a settlement [7], while a denial may suggest that individualized lawsuits or new legislative measures could be the only viable paths forward [7].
As the case progresses into discovery and trial [3], a ruling in favor of the plaintiffs could reshape the framework for AI companies [3], necessitating a shift in how they document their data sources and licenses [3]. Companies may need to move away from vague claims of “publicly available data” to a more transparent approach [3], emphasizing robust opt-out mechanisms and consent-based data inclusion strategies [3]. The outcome of the interlocutory appeal by the Ninth Circuit will play a crucial role in shaping the proceedings and the evolving legal framework governing the use of copyrighted materials in AI training. These developments align with the US Supreme Court’s reasoning in Google v [5]. Oracle (2021) [5], where transformative use was pivotal in determining fair use regarding software APIs [5]. As AI technology progresses [5], the legal landscape governing its use will continue to evolve [5], influencing how judges [6], regulators [3] [6], and companies navigate copyright compliance in this rapidly changing field [6].
Conclusion
The ongoing legal battle in Bartz v. Anthropic underscores the complexities of copyright law in the context of AI development. The case’s outcome could have far-reaching effects on how AI companies handle copyrighted materials, potentially leading to stricter documentation and consent practices. The decision by the Ninth Circuit on the interlocutory appeal will be pivotal in shaping future legal standards and industry practices, reflecting broader trends in the intersection of technology and intellectual property law.
References
[1] https://chatgptiseatingtheworld.com/2025/08/05/anthropic-filed-rule-23f-petition-to-appeal-judge-alsups-class-certification-in-9th-circuit/
[2] https://tukcopyrightlaw.substack.com/p/authors-get-class-certification-in
[3] https://avenuez.com/blog/the-bartz-class-action/
[4] https://news.bloomberglaw.com/ip-law/anthropic-to-appeal-class-certification-in-ai-copyright-case
[5] https://www.escaladelegal.com/u-s-court-rules-ai-training-on-copyrighted-books-qualifies-as-fair-use-key-implications-for-tech-and-legal-sectors/
[6] https://airevolution.poltextlab.com/anthropic-spent-tens-of-millions-of-dollars-to-compile-unique-books-dataset-for-claude-ai-training/
[7] https://www.authorsalliance.org/2025/06/18/bartz-v-anthropic-class-certification-at-issue-for-book-authors-in-ai-copyright-litigation/
