The Third Circuit Court of Appeals delivered a decisive victory to Thomson Reuters in a copyright dispute with ROSS Intelligence, an artificial intelligence-powered legal research platform, ruling that Westlaw headnotes qualify for copyright protection and that ROSS did not establish a fair use defense for its use of that material.
The decision addresses a core tension in legal publishing. Thomson Reuters maintains that headnotes, the brief summaries preceding court opinions in Westlaw's database, represent original creative work deserving copyright protection. ROSS Intelligence, which uses machine learning to analyze and categorize legal opinions, argued that headnotes are factual summaries of public domain judicial opinions and therefore uncopyrightable. The startup also contended its use fell within fair use doctrine.
The Third Circuit rejected both arguments. The court found headnotes embody sufficient originality to merit copyright protection. These summaries require editorial judgment, analysis, and selection of relevant legal principles from sprawling opinions. Thomson Reuters' authors make creative choices about emphasis, scope, and framing that exceed bare factual compilation.
On fair use, the court applied the four-factor test from 17 U.S.C. Section 107. The panel examined the nature and purpose of ROSS's use, the character of the copyrighted work, the amount taken, and the effect on the market for Thomson Reuters' product. The Third Circuit found ROSS's wholesale incorporation of headnotes into its AI-training infrastructure transformed them into a competing commercial product. ROSS did not meaningfully add new value or commentary. Instead, the startup leveraged Thomson Reuters' labor to build market share in legal research.
The market harm prong proved decisive. Thomson Reuters licenses Westlaw content to law firms, corporations, and other customers. ROSS offers subscription-based legal research services that directly compete with Westlaw. By copying headnotes without permission or compensation, ROSS undermined Thomson Reuters' ability to monetize its copyrighted content. The Third Circuit found this commercial harm weighed heavily against fair use protection.
This decision carries substantial implications for legal publishing and the intersection of copyright law with artificial intelligence development. Legal research has historically involved copying and organizing existing materials, but the Third Circuit distinguished between fair use analysis and wholesale appropriation for competitive advantage. AI companies cannot simply train systems on protected content without licensing, even when underlying case law remains in the public domain.
The ruling also affects the broader debate about copyright's role in protecting editorial work. Thomson Reuters and competing publishers like LexisNexis invest heavily in creating value-added legal content. The Third Circuit's copyrightability finding protects that investment model. Publishers can continue licensing content to AI developers rather than watching those developers build competing products from copied materials.
ROSS Intelligence faces a decision on further appeal, though Supreme Court review seems unlikely given the relatively narrow copyright scope. More immediately, the ruling establishes that legal AI companies must negotiate licenses for headnotes and similar editorial content rather than assuming public domain case law permits copying everything around it.
The case reinforces that copyright protects creative editorial choices even when the underlying subject matter remains public. For legal technology companies, the decision signals that building competitive products requires either licensing copyrighted content or developing genuinely original analytical frameworks independent of existing publisher work.