Overview
- The publishers filed a 38‑page complaint on Friday in the U.S. District Court for the Southern District of New York, alleging OpenAI and Microsoft scraped hundreds of thousands of articles, including material behind paywalls, to build AI training datasets.
- The complaint cites examples of near‑verbatim reproductions by AI, including an 88‑word excerpt from The Seattle Times, and asks the court for monetary damages plus orders to impound or destroy copies of the plaintiffs’ works, datasets and models.
- OpenAI says its models are trained on publicly available data and invoke fair use, Microsoft said it was surprised and is willing to discuss solutions, and the Justice Department has filed a statement in a related New York Times case arguing that training on copyrighted material alone is not infringement.
- The suit joins a consolidated wave of publisher litigation in the Southern District of New York that includes The New York Times and roughly 400 local papers, positioning the court to set a precedent on transformation, licensing and developers’ dataset practices.
- Publishers warn of concrete economic harm, citing roughly a 47% drop in search referral traffic for midsized outlets and arguing AI answers reduce clicks and subscriptions, a shift that could prompt more licensing deals or changes in how models are trained.