Appeals court rules AI firm's copying of Westlaw content wasn't fair use
The Third Circuit upheld Thomson Reuters' win against Ross Intelligence, the first appellate ruling on whether copying content to train an AI system counts as fair use.
What happened: A federal appeals court, the Third Circuit, upheld Thomson Reuters' copyright victory over Ross Intelligence, a now-defunct legal research startup. The case centered on Westlaw headnotes, short summaries of legal points from court rulings, that Ross had used to train its own AI-powered legal search engine. The appeals court rejected Ross's argument that this counted as fair use. The court's full reasoning was sealed at the time it was reported, so the detailed legal logic isn't yet public.
Why it matters: Reuters described this as a first-of-its-kind appellate ruling in the wave of copyright lawsuits over AI training data. Courts have so far split on similar questions in other cases, and this is the first time an appeals court, rather than a lower trial court, has weighed in. That makes it a reference point other judges may look to, even though it doesn't automatically decide how other pending cases will go.
How it works, plainly: Thomson Reuters sued Ross Intelligence in 2020, alleging Ross copied Westlaw material to build a competing AI search tool. Ross argued its use of the headnotes was transformative enough to qualify as fair use under copyright law. Both the trial court and now the appeals court disagreed, siding with Thomson Reuters. The case is specifically about a legal search product, not a general-purpose chatbot, which is an important distinction reporters covering it have flagged.
The rollout: The practical fallout already happened years ago: Ross shut down its platform in January 2021, citing the cost of fighting the lawsuit. A separate antitrust countersuit Ross brought against Thomson Reuters, accusing it of illegally bundling Westlaw products, was dismissed in September 2024. Thomson Reuters said it believes respecting copyright is essential to building trustworthy AI tools, while more AI-training copyright cases, including those from news publishers against chatbot makers, remain pending and unresolved.
