An unsealed opinion from the 3rd U.S. Circuit Court of Appeals on Wednesday, September 30, 2026, confirmed that Ross Intelligence did not make fair use of Thomson Reuters' copyrighted material to develop an AI-based legal search engine. The unanimous three-judge panel affirmed a prior ruling from a Delaware district court, finding that the defunct company’s use of Westlaw "headnotes" to train a competing product was not transformative.
The legal dispute began in 2020 when Thomson Reuters, the parent company of Reuters News, sued Ross Intelligence for allegedly copying thousands of headnotes—short summaries of legal points in court opinions. Ross Intelligence argued that its use of these materials was protected under the fair use doctrine and that the headnotes themselves were not copyrightable. The company shut down its platform in 2021, citing the high costs associated with the litigation.
In the unsealed opinion, U.S. Circuit Judge Tamika Montgomery-Reeves wrote that the case was an "ordinary copyright case" rather than a broader referendum on the future of AI. The court concluded that Thomson Reuters’ materials possessed a "creative spark" and that Ross intended to be a direct competitor by using them for a highly similar purpose. The judges also distinguished this case from other high-profile AI training lawsuits, noting that Ross’ search engine did not utilize generative AI to create new content.
The scale of the dispute involved thousands of copyrighted headnotes produced by Thomson Reuters' attorney editors. For the defendant, Ross Intelligence, the impact of the lawsuit was total, leading to the 2021 closure of a business that once aimed to compete in the legal-research market. The court’s distinction between this case and those involving generative AI—like the ongoing litigation between the New York Times and OpenAI—clarifies that not all AI training methods will be treated identically under copyright law. Tech firms and small-business owners developing AI tools now face a narrower path for fair use defenses if their products directly compete with the original data source.
The ruling sets a precedent that editorial content maintained by humans remains protected even when used as training data for machine learning. A spokesperson for Thomson Reuters stated the company was pleased with the decision, emphasizing that strong copyright protections are essential for innovation. Legal representatives for Ross Intelligence did not immediately respond to requests for comment. Following this appellate affirmation, the case returns to the lower court's established findings, and the immediate legal path for Ross Intelligence to utilize these specific headnotes remains blocked.