Seattle Times and Newsday sue OpenAI and Microsoft over alleged use of copyrighted journalism in AI training

The lawsuit and its core allegations

The Seattle Times and Newsday filed a copyright complaint against OpenAI and Microsoft in the Southern District of New York on September 4, 2026, alleging that the companies scraped their websites, including paywalled content, and incorporated copies into datasets used to train, fine-tune and ground AI models. The complaint names WebText, WebText2, Common Crawl and Microsoft's Bing search index among the sources allegedly used, and says the publishers' content was copied despite terms-of-service restrictions. Seattle Times prohibits using its content for AI training or grounding and automated scraping, while Newsday's terms prohibit using its content to develop software, including AI systems; Newsday also says its robots.txt file instructed OpenAI and Common Crawl not to crawl its site. The publishers further allege that the defendants removed copyright information, including authors' names, article titles and copyright notices, from some copies.

The publishers' central legal argument

The publishers argue that the dispute is not simply whether AI models can learn from publicly available material, but that OpenAI and Microsoft copied protected journalism and used those copies to build commercial products that can provide readers with the same information without directing them to the original publishers. The complaint puts the point bluntly: "There is nothing transformative about copying The Seattle Times' and Newsday's journalism …" It also alleges that AI systems can reproduce their reporting almost word for word, citing a test in which a model allegedly reproduced 88 consecutive words from a Seattle Times article about the Boeing 737 MAX crisis when given the article's headline and URL, alongside further examples involving Newsday articles. The complaint describes the broader concern by asserting that "AI products like ChatGPT and CoPilot are touted as producers of content, but in fact they are rapacious consumers," and warns that "AI that is trained on painstakingly researched, expensive-to-produce content threatens to destroy the very news organizations by competing directly with them."

The licensing market the plaintiffs cite

The publishers point to OpenAI's existing licensing agreements with the Associated Press, News Corp, Axios, Axel Springer, The Atlantic, Financial Times, Dotdash Meredith and Vox Media as evidence that a commercial market exists for licensing news content to AI systems, and they allege that OpenAI never sought or obtained a license from either plaintiff. The complaint says publicly disclosed terms of three agreements show OpenAI paid more than $300 million for news-content rights, while the terms of the other agreements remain private.

The relief and damages the publishers are seeking

The publishers say AI-generated answers can reduce visits to their websites, affecting advertising and subscription revenue, and they are pursuing monetary damages while also seeking the destruction of any training datasets and AI models that incorporate their copyrighted content. By naming Microsoft as a co-defendant, the plaintiffs are targeting the infrastructure behind Copilot, which relies heavily on OpenAI's technology. The lawsuit aligns the two outlets with a broader coalition of nearly 400 local newspapers that are pursuing similar legal action, alongside other media organizations including The New York Times, Ziff Davis and Merriam-Webster that have alleged their intellectual property was misappropriated.

Where the case sits alongside prior litigation

The new complaint adds to OpenAI and Microsoft's existing copyright exposure, following The New York Times' separate suit, and the defendants have publicly argued that training and operating AI on news content constitutes fair use. The plaintiffs contest that framing by emphasizing verbatim reproduction and the removal of copyright information, allegations that, if proven, would distinguish the conduct from the kind of transformative learning the defendants invoke. The dispute now centers on whether the scraping of paywalled and terms-restricted journalism, and the verbatim output of that journalism by AI products, crosses the line from fair use into infringement under the publishers' pleaded theories.