Introduction

The recent ruling by Judge Stephanos Bibas in the District of Delaware has significant implications for copyright law, particularly concerning the use of copyrighted materials by AI companies. This case involves a copyright infringement lawsuit filed by Thomson Reuters against ROSS Intelligence, an AI search tool developer [4], focusing on the fair use of copyrighted headnotes.

Description

Concord Music aims to emphasize Judge Stephanos Bibas’s recent ruling in the District of Delaware, which updates a previous summary judgment in the copyright infringement lawsuit filed by Thomson Reuters against ROSS Intelligence [4], an AI search tool developer [4]. The judge’s decision revisits the fair use analysis [4], highlighting that ROSS’s use of Thomson Reuters’ copyrighted headnotes—specifically 2,243 headnotes that summarize legal decisions—was deemed non-transformative and did not qualify as “fair use” under copyright law.

In his updated ruling [4], Judge Bibas acknowledged that his earlier decision from September 2023 did not fully address the complexities of the case. While that prior ruling had largely dismissed motions from both parties, determining that many disputed issues [4], including the validity of Thomson Reuters’ copyright in its headnotes and ROSS’s fair use defense [4], needed to be resolved by a jury [4], the recent ruling granted most of Thomson Reuters’ motion for partial summary judgment on copyright infringement and fair use [4]. The judge denied ROSS’s motion for summary judgment on fair use, emphasizing the significance of legal research platforms and noting Westlaw’s prominence in accessing legal materials [4].

The court’s analysis of the four statutory factors of fair use revealed that, while two factors favored ROSS [2], critical aspects regarding the purpose of the use and its potential harm to the market for the original content favored Thomson Reuters. The judge found strong circumstantial evidence of copying [3], noting that the Bulk Memos produced by ROSS resembled headnotes more than the underlying judicial opinions [3]. He concluded that ROSS’s use was commercial and non-transformative [3], lacking a further purpose beyond that of Thomson Reuters [3], and emphasized that ROSS intended to compete with Thomson Reuters [3], negatively impacting the value of the copyrighted work [7].

This ruling partially granted summary judgment on Thomson Reuters’ copyright infringement claims [1] [2], affirming the originality of Westlaw’s headnotes for copyright protection, although it left unresolved whether copyright protection has expired for any of them [1]. The court distinguished this case from others involving intermediate copying of computer source code [3], where such copying was deemed necessary for reverse engineering [3]. In this instance [3], the court concluded that copying Thomson Reuters’ protected expression was not essential to access the underlying ideas [3].

Judge Lee will be presented with this information in an ongoing case, despite opposition from Anthropic regarding the motion for leave [5]. While Judge Lee will review the matter [5], her assessment of the reasoning’s persuasiveness remains uncertain [5]. The ruling also dismissed ROSS’s other defenses, including claims of innocent infringement and copyright misuse [2], reflecting a shift from Judge Bibas’s earlier decision in 2023 [7], where he did not issue a summary judgment on the fair use claim due to precedents involving intermediate copying of computer programs [7].

Although ROSS ceased operations in January 2021 due to litigation costs [2] [7], the case continues as Thomson Reuters seeks damages for the alleged infringement [7]. Remaining issues for trial include whether ROSS copied the Key Number System and certain court opinions [2], as well as questions of contributory and vicarious liability [2]. This ruling is significant as one of the first copyright lawsuits addressing AI product development and may have substantial implications for AI companies in the legal sector regarding the use of copyrighted materials for training AI systems that compete with existing products. Intellectual property experts anticipate that this ruling will be pivotal for artists and content creators in their legal battles against generative AI systems [6], reinforcing the notion that using copyrighted material as training data does not automatically constitute fair use [6]. Following this decision [1], music publishers involved in a separate case against Anthropic PBC referenced the ruling in their ongoing litigation concerning the use of copyrighted song lyrics for AI training [1]. Thomson Reuters expressed satisfaction with the court’s decision [1], which underscores the copyright protection of its editorial content and criticized proposed government measures that could erode copyright protections for AI companies.

Conclusion

The ruling by Judge Bibas is a landmark decision in the realm of copyright law, particularly for AI companies. It underscores the importance of respecting copyright protections and clarifies the boundaries of fair use in the context of AI development. This decision is likely to influence future legal battles involving AI and copyrighted materials, setting a precedent for how such cases are approached and resolved.

References

[1] https://news.bloomberglaw.com/ip-law/judge-rejects-fair-use-defense-in-westlaw-ai-copyright-lawsuit
[2] https://www.lawnext.com/2025/02/breaking-federal-judge-rules-legal-research-startup-ross-infringed-westlaws-copyrights-rejecting-fair-use-defense.html
[3] https://natlawreview.com/article/court-training-ai-model-based-copyrighted-data-not-fair-use-matter-law
[4] https://ipwatchdog.com/2025/02/12/revised-fair-use-ruling-finds-no-transformative-use-developing-ai-search-tool/id=185989/
[5] https://chatgptiseatingtheworld.com/2025/02/12/concord-music-wants-to-cite-to-judge-bibass-rejection-of-fair-use/
[6] https://www.hollywoodreporter.com/business/business-news/reuters-ross-first-ai-copyright-case-lawsuit-1236133743/
[7] https://completemusicupdate.com/us-judge-demolishes-fair-use-defence-in-what-could-be-landmark-ai-lawsuit/