The Seattle Times Company and Newsday LLC have sued OpenAI and Microsoft in federal court in Manhattan, alleging that the companies copied the publishers’ journalism without permission for use in generative-AI systems. The complaint was filed September 4, 2026, in the U.S. District Court for the Southern District of New York.
The publishers are seeking damages and court-ordered relief over what they characterize as unauthorized use of their work. Their lawsuit adds two regional news organizations to the growing group of publishers challenging whether AI companies may copy journalism without a license to train, operate or improve large language models and the products built around them. Reuters reported that the complaint accuses OpenAI and Microsoft of copying the newspapers’ journalism to train AI systems. (Investing.com)
What the complaint alleges
The complaint alleges that the defendants acquired Seattle Times and Newsday articles through automated web scraping, including material that the publishers say was available only behind their paywalls. It contends that the articles were incorporated into datasets and used to train and operate the language models associated with ChatGPT, Microsoft Copilot and Bing AI features. Those assertions are allegations in a newly filed civil complaint, not findings of fact or liability by a court.
The publishers’ theory extends beyond the initial acquisition of articles. They contend that the AI products can generate verbatim passages or close paraphrases of protected reporting, potentially supplying information that a reader otherwise would obtain from the publishers’ sites. The complaint cites an example in which ChatGPT allegedly returned 88 consecutive words from Seattle Times reporting, according to accounts describing the filing.
The lawsuit includes copyright claims and trademark-related claims, including allegations that AI-generated material has at times been falsely associated with the publishers’ brands. Each of those contentions remains to be tested through the litigation process.
Requested relief and the companies’ responses
The relief requested is broad. In addition to damages, the publishers ask the court to order the destruction or impoundment of copies of their works and of datasets or models that incorporate them. Such requests do not mean that a court will impose those remedies: the complaint states the plaintiffs’ requested outcome, while the defendants will have an opportunity to contest both liability and the scope of any available relief.
Microsoft said it was surprised by the lawsuit, recognized the importance of local journalism and was open to exploring solutions to this kind of dispute. OpenAI did not specifically address the Seattle Times-Newsday complaint in the Reuters report, but said its models are trained on publicly available data and that its practices are grounded in fair use. (Investing.com)
The broader fair-use debate
The fair-use question sits at the center of this and similar cases. The publishers argue that copying their work is unlawful, particularly where AI products allegedly reproduce or closely track reporting and can function as substitutes for visiting the original source. AI developers have generally maintained that training on large collections of publicly available material is a legally protected transformative use. In the separate New York Times litigation, the U.S. Justice Department recently supported OpenAI’s position that training large language models on vast collections of writing can qualify as fair use, underscoring the policy importance of the issue beyond this one dispute. (AP News)
The Seattle Times and Newsday action is separate from The New York Times’ 2023 lawsuit against OpenAI and Microsoft, though it raises related questions about training data, model outputs and the economic value of original reporting. A federal judge previously allowed key copyright claims in the Times case and related newspaper claims to proceed while dismissing or narrowing some other claims, illustrating that courts are beginning to sort through the competing legal theories without yet producing a definitive rule for generative-AI training. (AP News)
For publishers, the case concerns both the alleged copying of journalism and the possibility that AI-generated answers may reduce the need for users to visit the source that funded the reporting. For AI companies, it is another test of whether existing copyright doctrines can accommodate web-scale model development. The outcome will depend on the evidence developed in this case and on legal rulings that remain ahead; the September 4 filing itself does not resolve whether OpenAI or Microsoft is liable.
