Key developments: A U.S. Third Circuit appeals court affirmed that Ross Intelligence’s use of Thomson Reuters’s Westlaw headnotes to train its legal‑research AI was not fair use, marking the first federal appellate decision directly addressing AI‑training copyright claims. The ruling emphasizes that traditional fair‑use analysis applies even to AI models, and it highlights the court’s focus on the copyrighted nature of the headnotes and the competitive purpose of Ross’s product.
Why it matters: The decision signals to AI developers that using copyrighted material without permission—even for non‑generative tools—carries significant legal risk. It provides a precedent that may shape future disputes over large language model training, where plaintiffs argue that massive data scraping violates copyright.
What to watch next: Publishers are intensifying the fight. USA Today and 13 of its newspapers have sued OpenAI, demanding more than $250 million for alleged copying of hundreds of thousands of articles, and accusing the company of harming traffic with AI‑generated summaries. The New York Times and other news outlets are pursuing similar claims against OpenAI and Microsoft. Separate cases, such as Concord Music Group’s suit against Anthropic over song‑lyric training data, are also pending. These lawsuits, combined with the Ross ruling, set the stage for a wave of litigation that could force AI firms to negotiate licensing deals or face costly judgments. [[1, 2]]