USA Today Co., the newspaper chain formerly known as Gannett, sued OpenAI for more than $250M in New York federal court, alleging it took articles by the hundreds of thousands across 19 of its papers to train ChatGPT and then served that reporting back to users. The complaint says the papers’ work appears in more than 160,000 entries of the dataset OpenAI built for GPT-2, and that GPT-5.6 pulled up and summarized Indianapolis Star and Detroit Free Press stories on request. USA Today Co. argues the copying was willful, pointing to OpenAI’s paid licensing deals with other publishers, and it wants the court to order OpenAI to destroy models and training data built on its content. USA Today Co. already licenses its own articles to Perplexity under a deal it signed last year.
