Introduction
The ongoing legal dispute between Kadrey and Meta highlights the complexities of copyright law as it intersects with the development of large language models (LLMs). The case centers on the use of copyrighted materials for AI training, with significant implications for the future of licensing and innovation in the tech industry.
Description
Kadrey and Meta are in disagreement regarding the next steps following Judge Vince Chhabria’s partial summary judgment in a copyright case involving authors and the training of large language models (LLMs). The judge ruled that Meta’s use of copyrighted books to train its Llama models constitutes fair use [1] [4], granting summary judgment on the copyright claim related to AI training [5]. This ruling applies specifically to the plaintiffs involved [4], including notable authors such as Sarah Silverman and Junot Díaz [1]. Kadrey is seeking supplemental briefing concerning Meta’s alleged uploading of pirated book copies [5], while Meta is requesting a partial final judgment on the issues already decided [5], aiming to appeal to the 9th Circuit and stay the outstanding claim [5].
The court found that Meta’s use of the copyrighted works was highly transformative, as the LLMs serve functions distinct from the original texts [3]. Judge Chhabria emphasized that the plaintiffs’ argument regarding Meta’s downloading from shadow libraries as automatic piracy was flawed, noting that fair use analysis focuses on the legality of copying actions rather than the source of the materials. While the use of shadow libraries could indicate bad faith [3], it did not significantly impact the case’s outcome [3]. The plaintiffs failed to provide evidence that Meta’s actions benefited shadow libraries or contributed to unauthorized copying [3], and the court determined that the amount of material copied was not particularly relevant [3], as the LLMs did not produce substantial outputs of the plaintiffs’ texts [3].
The remaining issue in the lawsuit pertains to the infringement claim regarding Meta’s alleged uploading of pirated books during the torrenting process [5]. Meta had initially sought to negotiate licensing deals but resorted to downloading books from shadow libraries after discovering that publishers typically do not hold subsidiary rights for AI training [1]. Kadrey wishes to advance the case in district court regarding the outstanding uploading claim [5], proposing supplemental briefing in August 2025 [5]. Conversely [5], Meta contends that Kadrey initially supported the idea of a partial final judgment but later altered their stance [5], a claim that Kadrey disputes [5]. Both parties have submitted statements on how they believe the judge should proceed [5], with a potential hearing set for August 28, 2025 [5].
Both parties agree that class certification [5], if applicable [5], should take place after the court resolves the outstanding uploading claim [5]. This ongoing legal battle highlights the evolving landscape of copyright law as it intersects with AI development, underscoring the need for proactive licensing arrangements with content creators [4]. Judge Chhabria criticized the notion that requiring licenses would stifle innovation [2], asserting that tech companies must compensate for the works they utilize [2]. This sets the stage for an increase in collective rights organizations and licensing frameworks aimed at monetizing creative contributions to AI systems [2].
Concerns were raised about whether AI-generated content competes with the plaintiffs’ works [3], particularly as generative models could dilute or displace commercial markets [2], which could significantly impact litigation outcomes [2]. However, the plaintiffs did not analyze their book markets or the potential impact of AI-generated books [3]. The judge’s analysis suggested that licensing payments might not effectively support individual authors [3], potentially benefiting larger content owners instead [3]. The legal landscape surrounding AI copyright liability is evolving [2], with recent cases underscoring the fragility of fair use defenses when plaintiffs present compelling arguments [2].
The intersection of strategic litigation choices and market impact is becoming increasingly critical [2], as fair use protections for AI developers are no longer guaranteed and must be substantiated through evidence and design considerations [2]. The focus is shifting from mere competition and similarity to the unauthorized use of copyrighted works [2]. The courts are recognizing that misappropriation [2], rather than competition [2] [3], is the core issue at hand [2]. The rulings in this case signal a departure from the assumption that fair use would provide blanket protection for AI training on copyrighted content [2], indicating a more nuanced legal environment where specific evidence and strategic choices will dictate outcomes [2].
Future developments will likely unfold in licensing negotiations and corporate strategies [2], with successful companies recognizing the importance of sustainable relationships with content creators [2]. The competitive landscape for AI development is now global [2], with national security implications [2], necessitating a balance between fostering innovation and respecting creators’ rights [2]. The ongoing legal discourse will shape the future of AI and copyright [2], emphasizing the need for reasonable licensing frameworks and clear legal standards that support both technological progress and the rights of creators [2]. The transition from litigation to collaboration among stakeholders will be crucial in navigating this new landscape [2].
Conclusion
The case between Kadrey and Meta underscores the evolving nature of copyright law in the context of AI development. It highlights the need for clear legal standards and proactive licensing arrangements to balance innovation with the rights of content creators. As the legal landscape shifts, the focus will be on sustainable relationships and strategic choices that respect both technological progress and intellectual property rights.
References
[1] https://ppc.land/court-rules-meta-used-copyrighted-books-legally-for-ai-training/
[2] https://aiforlawyers.substack.com/p/fair-use-on-trial-what-two-ai-decisions
[3] https://www.authorsalliance.org/2025/06/26/meta-wins-on-fair-use-for-now-but-court-leaves-door-open-for-market-dilution/
[4] https://fusionchat.ai/news/ai-vs-copyright-metas-legal-battle-over-training-data
[5] https://chatgptiseatingtheworld.com/2025/07/09/kadrey-meta-disagree-on-how-to-proceed-after-judge-chhabrias-partial-summary-judgment-kadrey-wants-discovery-and-supplemental-briefing-on-metas-alleged-uploading-of-book-copies-meta-wants-parti/
