September 7, 2026 · Legal
The Seattle Times Co. and Newsday filed federal copyright and trademark claims against OpenAI and Microsoft, alleging the companies trained AI models on their journalism without permission or payment. GeekWire and TechCrunch reported the suit seeks damages and an injunction. One account said the publishers want models trained on their articles destroyed.
The filing joins a crowded docket. The New York Times is already in a closely watched case against the same defendants; recent court papers in that suit leaned on copyright doctrine plus cultural and sports references. Local papers suing a hometown tech giant—Microsoft sits in the Seattle Times’ own metro area—adds a political edge the national cases lack.
Publishers argue that generative models are derivative works built from newsrooms that still pay reporters. Labs argue that training is fair use and that licensing every article at internet scale is unworkable. Courts have not settled the question, which is why each new plaintiff matters: more venues, more facts, more pressure to settle or to set a precedent.
Whatever the legal theory, the business implication is immediate. If injunctions that force model deletion become plausible, training data provenance stops being a compliance footnote and becomes a balance-sheet risk for every lab that scraped the open web.
Key takeaway. Two major regional publishers are asking a federal court not just for money, but for models trained on their work to be taken down. That remedy, if it ever sticks, would reshape how labs treat news archives.
Photo: Unsplash (photo-1504711434969-e33886168f5c). Sources: GeekWire; TechCrunch / AInformed; The New York Times; AI Weekly.
