Introduction
The ongoing legal case of Bartz v Anthropic PBC, presided over by Judge William Alsup, examines whether the use of pirated books for training AI models by Anthropic PBC constitutes fair use under copyright law. This case has significant implications for the AI industry, copyright holders [2], and the legal frameworks governing intellectual property.
Description
Judge William Alsup is currently evaluating whether Anthropic PBC’s use of pirated books for training AI models constitutes fair use under copyright law in the case of Bartz v Anthropic PBC. He appears inclined to accept fair use [5], although he remains open to changing his position [5]. During oral arguments [5], signs indicate that he recognizes a transformative purpose in AI training [5], as he directed Anthropic’s attorney to address additional factors of fair use and challenged the opposing counsel’s comparisons to Napster [5]. Notably, there was no inquiry into the Copyright Office’s pre-publication report referenced by the opposing party [5], although plaintiffs Kadrey and Bartz have cited this report as Supplemental Authority in their lawsuits against both Meta and Anthropic. This report [4], which is generally favorable to the plaintiffs regarding fair use issues [4], has raised concerns about the authority of the Copyright Office following the termination of Register of Copyrights Shira Perlmutter [4], as the new Register may have differing views or could potentially rescind the report [4].
Anthropic argues that its use of copyrighted materials is transformative [1], favoring the fair use analysis based on the four factors: purpose [1], nature [1] [2], amount used [2], and market effects [1]. The case raises critical questions about the legality of utilizing shadow libraries—unauthorized repositories of copyrighted material—for AI training [2], with authors’ counsel contending that Anthropic’s reliance on these networks undermines its fair use defense [1], as they are detrimental to the authors’ market [1]. Judge Alsup has raised questions about the authors’ claims regarding AI’s market impact [1], particularly concerning the potential dilution of the market by AI-generated content [1]. He has requested further analysis of the 2015 Second Circuit decision in Authors Guild v Google Inc [1], which explored the complexities of copyright in digital contexts and may provide relevant insights into fair use protections for original copying of copyrighted works. Additionally, Judge Alsup has requested further briefing on the fair use implications of Google’s book copying in the context of the Authors Guild case [5].
In a related development, a federal judge has ordered Anthropic to address allegations regarding a false citation included in a court document related to a copyright infringement case involving music publishers [3]. The lawsuit claims that Anthropic improperly used copyrighted lyrics to train its AI chatbot [3], Claude [3]. An expert testimony submitted by Anthropic cited a non-existent academic article [3], which was identified as a fabrication by the plaintiffs’ attorney [3], who suggested that the expert may have relied on the AI for generating the content [3]. The judge highlighted the seriousness of the issue [3], distinguishing between a simple citation error and a potentially AI-generated false reference [3]. Anthropic’s attorney acknowledged the citation was incorrect but argued it likely referred to a real article [3]. This case reflects the increasing scrutiny of AI-assisted legal work and the challenges faced by AI developers regarding the use of copyrighted material [3].
The outcome of the current case could significantly reshape the legal landscape for AI companies that depend on large datasets, some of which may be pirated [2]. A ruling against Anthropic could compel AI companies to secure licenses for training data [2], significantly impacting operational costs and innovation [2], particularly for smaller firms [2]. Conversely [2], a ruling in favor of Anthropic might challenge existing copyright structures [2], potentially undermining the rights of original content creators [2]. The implications of this case extend beyond the immediate parties involved [2], influencing the broader AI community and copyright holders [2]. The ongoing legal discourse highlights the tension between technological advancement and intellectual property protection [2], emphasizing the need for updated legal frameworks that address the unique challenges posed by AI [2]. The outcome may set a precedent for future interactions between AI technologies and copyright law [2], underscoring the importance of balancing innovation with the rights of creators [2].
Conclusion
The case of Bartz v Anthropic PBC is pivotal in determining the future of AI development and copyright law. Its outcome could redefine the boundaries of fair use, influence the operational practices of AI companies, and impact the rights of content creators. As the legal proceedings unfold, the case underscores the necessity for evolving legal frameworks that can accommodate the rapid advancements in AI technology while safeguarding intellectual property rights.
References
[1] https://news.bloomberglaw.com/ip-law/judge-hints-anthropics-ai-training-on-authors-work-is-fair-use-62
[2] https://opentools.ai/news/anthropics-ai-copyright-showdown-fair-use-in-the-spotlight
[3] https://www.pymnts.com/cpi-posts/anthropic-ordered-to-respond-after-ai-allegedly-fabricates-citation-in-legal-filing/
[4] https://chatgptiseatingtheworld.com/2025/05/13/with-summary-judgment-pending-kadrey-bartz-plaintiffs-cite-pre-publication-version-of-copyright-office-report-on-fair-use-and-ai-training-as-supplemental-authority-despite-its-non-final-status/
[5] https://chatgptiseatingtheworld.com/2025/05/22/judge-alsup-says-hes-leaning-to-accept-fair-use-by-anthropic-but-cautions-may-change-his-mind/
