Individual case record
Thomson Reuters Enterprise Centre GmbH v. ROSS Intelligence Inc.
Cause: 17:101 Copyright Infringement
- Court
- D. Del.
- Docket
- 1:20-cv-00613
- Filed
- 6 May 2020
- AI system
- AI-powered legal search engine
Individual case record
Overview
Filed in December 2020, this is the first case in which a US court has substantively rejected a fair use defence for AI training. Thomson Reuters sued AI startup ROSS Intelligence for using Westlaw's copyrighted headnotes and Key Number System to train a competing AI-powered legal search engine. On 11 February 2025, the court granted partial summary judgment for Thomson Reuters, finding that ROSS's use was commercial, not transformative, and threatened the potential market for licensing AI training data. The case is now on interlocutory appeal to the Third Circuit.
Facts & Claims
Filed in December 2020 in the District of Delaware, this action arose from ROSS Intelligence's development of an AI-powered legal search engine designed to compete with Thomson Reuters's Westlaw platform. ROSS's tool was not generative AI; it was designed to return ranked passages from existing judicial opinions in response to natural language queries, rather than to generate new text.
ROSS had initially approached Thomson Reuters about licensing Westlaw content to train its AI tool, but Thomson Reuters declined. ROSS then engaged a third-party company, LegalEase Solutions, to create approximately 25,000 "bulk memos" summarising legal issues and identifying relevant cases. LegalEase created these memos using Westlaw's resources, including its headnotes and Key Number System. The resulting bulk memo questions closely tracked the language of the Westlaw headnotes. ROSS used the bulk memos as training data for its search engine. Thomson Reuters alleged direct copyright infringement, claiming that ROSS had unlawfully copied its headnotes and Key Number System. ROSS raised several defences, principally fair use, and argued that the headnotes lacked sufficient originality to qualify for copyright protection.
In September 2023, Judge Bibas largely denied both parties' cross-motions for summary judgment, finding genuine factual disputes on copyrightability and fair use. However, after inviting renewed briefing, Judge Bibas reversed course on 11 February 2025, granting partial summary judgment to Thomson Reuters on both direct infringement and fair use.
On copyrightability, the court held that Westlaw's headnotes, both individually and as a compilation, met the originality threshold under Feist, finding that they "introduce creativity by distilling, synthesising, or explaining part of an opinion." Of 2,830 headnotes examined, 2,243 were found to have been actually copied and substantially similar to the bulk memo questions.
On fair use, the court rejected ROSS's defence. On the first factor (purpose and character of the use), applying the Supreme Court's framework from Andy Warhol Foundation v. Goldsmith (2023), the court held that ROSS's use was not transformative because it did not have a "further purpose or different character" from Thomson Reuters's: both companies created legal research tools designed to identify relevant points of law. The court distinguished the intermediate copying cases involving reverse engineering of computer code (Sega, Connectix) on the basis that ROSS had copied written expression, not functional code. On the fourth factor (effect on the market), the court identified ROSS's tool as a potential competitor in both the legal research platform market and the emerging market for licensing data for AI training. Factors two and three favoured ROSS, but the court held that factors one and four, being more heavily weighted, were dispositive.
On 4 April 2025, Judge Bibas stayed the trial and certified two questions for interlocutory appeal: whether Westlaw's headnotes and Key Number System are sufficiently original for copyright protection, and whether ROSS's use constituted fair use. The Third Circuit accepted the appeal on 17 June 2025, making it the first appellate court to grant review of a copyright/AI case. ROSS filed its opening brief in September 2025, Thomson Reuters responded in November 2025, and ROSS's reply brief was filed in December 2025. Numerous amici have filed briefs on both sides, including the Electronic Frontier Foundation, the Authors Alliance, the Association of American Publishers, the Copyright Alliance, and law professors Edward Lee, Matthew Sag, Pamela Samuelson, and Rebecca Tushnet.
Both Judge Alsup in Bartz v. Anthropic and Judge Chhabria in Kadrey v. Meta distinguished this case in their June 2025 fair use rulings, noting that ROSS's tool was not generative AI and that it was used to create a directly competing product rather than to repurpose training data for a different type of use.
A hearing on the merits took place in June 2026.
This summary is provided for educational and research purposes only and does not constitute legal advice. Allegations described are those made in the filings. Please consult the underlying source documents.
Source documents
Documents
Read the complaint and available amendments. For the latest filings, view the source docket ↗.
Loading document records…
Connections
Related content
No approved public connections are available for this dispute yet.
Procedural record
Court & docket history
Docket entries from the past three months. For the complete history and latest filings, follow the source docket.
View full docket on CourtListener ↗Loading docket entries…
Latest filing recorded: 26 Jun 2025.