Introduction
The legal dispute between Ross Intelligence and Thomson Reuters centers on the copyright implications of using Westlaw headnotes in AI training. This case highlights significant issues regarding copyright law’s application to artificial intelligence, particularly concerning the concepts of fair use and originality.
Description
Ross Intelligence has been granted certification for an interlocutory appeal regarding a pivotal court ruling by Judge Stephanos Bibas of the US District Court for the District of Delaware. This ruling determined that the use of Westlaw headnotes—concise summaries of legal points created by Thomson Reuters editors—by Ross’s AI-powered legal tool constitutes copyright infringement. The court emphasized that Ross’s use of Thomson Reuters’s legal resources is deemed commercial rather than transformative [6], lacking a “further purpose or different character.” The ruling found that Ross had copied 2,243 headnotes, with the Bulk Memos resembling the headnotes more than the underlying judicial opinions [1]. This decision marks the first summary judgment to reject a fair use defense in an AI-training copyright case and could significantly impact how AI companies acquire and utilize training data, necessitating a reevaluation of development practices across the industry [1]. The court also highlighted the significance of market impact, noting that Ross did not provide adequate evidence to demonstrate that potential AI training markets would be unaffected [6], and concluded that Ross’s actions created a direct market substitute for Westlaw’s offerings [1].
In its appeal, Ross challenges the district court’s decisions on the fair use and copyrightability of Thomson Reuters’ content [3], arguing that the case raises urgent questions about AI and copyright law [4]. The company asserts that the legal theories involved downplay the significance of originality and overemphasize copyright protection [5], which it believes stifles AI innovation by hindering “fair learning” from factual statements [5]. Ross seeks appellate review on two key legal questions: whether Westlaw’s headnotes meet the originality requirement of the Copyright Act [4] [5], particularly given their lack of a “creative spark,” and whether Ross’s use of 0.076% of these headnotes to train an AI search engine constitutes transformative use or fair use [5]. The company contends that these issues are controlling questions of law with substantial grounds for differing opinions [4] [5], citing conflicting opinions from the court itself in previous rulings as evidence of this.
Thomson Reuters [1] [2] [3] [4] [5] [6] [7], in opposition to the appeal, argues that addressing these issues requires a fact-intensive analysis that is unsuitable for an appellate court [3]. This decision serves as a cautionary note for AI developers regarding the use of copyrighted materials in training AI systems [1], indicating that such use can constitute infringement even if the final product does not display the copyrighted content [1]. Judge Bibas clarified that the ruling pertains only to non-generative AI [1], leaving unresolved questions about copyright law’s application to generative AI systems [1]. Further factual determinations regarding specific headnotes and issues of contributory liability and damages will be addressed at trial [1], marking a significant judicial interpretation at the intersection of copyright law and artificial intelligence [1], potentially influencing future practices in AI training data acquisition [1]. It was also clarified that the public does not possess a right to access Thomson Reuters’ copyrighted materials [6], which are protected to encourage the creation of original works [6]. The court’s ruling underscores critical issues surrounding copyright infringement in AI, particularly regarding the legality of training data and the implications for the development of legal technology [7].
Conclusion
The outcome of this case could have far-reaching implications for the AI industry, particularly in how companies approach the acquisition and use of training data. It underscores the need for careful consideration of copyright laws in AI development and may prompt a reevaluation of current practices to ensure compliance and foster innovation within legal boundaries.
References
[1] https://sunslegal.com/2025/03/26/from-headnotes-to-headlines-the-reuters-vs-ross-intelligence-case/
[2] https://news.bloomberglaw.com/litigation/judge-allows-midstream-appeal-of-westlaw-ai-fair-use-decision
[3] https://news.bloomberglaw.com/artificial-intelligence/reuters-hits-back-at-bid-for-quick-appeal-of-ai-fair-use-win
[4] https://www.jdsupra.com/legalnews/court-grants-interlocutory-appeal-on-ai-6692484/
[5] https://natlawreview.com/article/court-grants-interlocutory-appeal-ai-fair-use-issue
[6] https://www.lexology.com/library/detail.aspx?g=45aaaeaf-fea6-40ce-bc7e-88d5ca8e8afa
[7] https://www.hlk-ip.com/news-and-insights/ross-cant-get-a-break-as-thomson-reuters-lays-down-the-law/
