3rd Circuit Rules For Thomson Reuters In Its Copyright Fight Against Legal Research Startup ROSS

The 3rd U.S. Circuit Court of Appeals ruled in favor of Thomson Reuters on the two issues before it in its copyright suit against shuttered AI legal research startup ROSS Intelligence, affirming Judge Stephanos Bibas's February 2025 summary judgment that Westlaw headnotes are copyrightable and that ROSS's use of them as AI training data was not fair use. The opinion remains under seal while the parties may request redactions, and it is the first federal appellate decision to address fair use in the context of AI training, though the copying predates generative AI and was done to build a product competing directly with Thomson Reuters. A Thomson Reuters spokesperson told Reuters the company was pleased and "firmly believes that respecting copyright is essential for fostering innovation while protecting the intellectual property that fuels fiduciary-grade AI solutions.

Yesterday I noted that th 3rd U.S. Circuit Court of Appeals had issued an opinion in the long-running copyright battle between Thomson Reuters and the now-shuttered AI legal research company ROSS Intelligence, but that the opinion was sealed and not viewable by the public. Today, it was revealed that the court has ruled in favor of Thomson Reuters on the two primary issues that were before it — copyrightability of the Westlaw headnotes and fair use. That makes it the first federal appellate decision to address fair use in the context of AI training, although the unique facts of the case, including that the copying was done before the advent of generative AI, may minimize the implications. While the opinion remains under seal while the court allows the parties to request redactions, what we know is that the 3rd Cicuit panel entered judgment affirming Judge Stephanos Bibas’s February 2025 summary judgment in favor of Thomson Reuters. For full coverage and background on this case, see my archive of stories https://www.lawnext.com/category/thomson-reuters-v-ross . The two issues on which the lower court had granted summary judgment were: - Whether a short quote or paraphrase of a judicial holding can be copyrightable. - Whether fair use protected ROSS’s internal use of Westlaw headnotes as training data for an AI legal-search engine whose outputs themselves did not infringe. Assuming the court’s opinion, when it is eventually release, is an unqualified affirmance, then that would leave Bibas’s rulings on both issues in place — that the Westlaw headnotes were sufficiently original to be protected and that ROSS’s use of that material to train its system was not fair use. While the ruling is unquestionably significant to the parties, the broader significance remains to be seen. Some have said that the case could have implications for other litigation involving OpenAI, Anthropic and Meta and their use of copyrighted materials to train their models. But ROSS’s use of TR’s materials happened before the widespread popularity of gen AI and specifically was for a non-generative AI system. Also, it was done, as Judge Bibas found, to build a legal research product designed to directly compete with Thomson Reuters. That is why it will be important to see the full opinion and read the court’s reasoning. An affirmance that closely follows the facts could be consequential but fairly confined. But an opinion that announces a broader rule could reverberate more widely. It is also worth noting that this was an interlocutory appeal from a summary judgment order. A Thomson Reuters spokesperson told Reuters https://www.claimsjournal.com/news/national/2026/09/30/340457.htm that the company was pleased with the ruling and “firmly believes that respecting copyright is essential for fostering innovation while protecting the intellectual property that fuels fiduciary-grade AI solutions.”