Seattle Times and Newsday Sue OpenAI and Microsoft, Demand AI Models Be Destroyed
Two more news outlets want courts to delete the AI systems trained on their journalism. The suits are part of a growing wave from nearly 400 local papers.
Key points
- The Seattle Times and Newsday filed copyright-infringement lawsuits against OpenAI and Microsoft in September 2026.
- Both outlets allege their articles were used without permission to train AI models, the software systems that power chatbots like ChatGPT.
- The publishers are asking courts to destroy not just copies of their work, but the training datasets and the AI models themselves.
- Microsoft is named because its Copilot assistant, a chatbot built into Microsoft products, runs on OpenAI's technology.
- Nearly 400 local newspapers have now sued the two companies over similar claims.
Two more newsrooms want a judge to erase the AI that learned from their work.
The Seattle Times and Newsday filed copyright-infringement suits against OpenAI and Microsoft in early September 2026, alleging that the companies scraped years of their published journalism to train AI models without asking and without paying. Both suits claim that when users ask those AI systems questions, the chatbots sometimes reproduce passages lifted almost word-for-word from the papers' original reporting.
Microsoft appears as a defendant because its Copilot product, an AI assistant built into Windows, Microsoft 365 and other software, is built on OpenAI's underlying technology.
What exactly are the publishers asking for?
They want more than money. The Seattle Times and Newsday are asking courts to order the destruction of any copies of their articles the companies are holding, the training datasets those articles fed into, and the AI models built from them.
That is a significant ask. Training datasets are enormous collections of text, sometimes hundreds of billions of words, used to teach an AI model how to write and reason. Deleting a model that has already been trained and deployed to millions of users would be a technical and commercial upheaval for OpenAI and Microsoft alike.
Neither company responded to a request for comment before publication.
How does this fit the bigger picture?
This is not an isolated complaint. These suits sit inside a much larger legal pile-up.
The New York Times sued OpenAI and Microsoft in late 2023. Ziff Davis, a large magazine publisher, filed its own suit. So did dictionary-makers Merriam-Webster and Encyclopedia Britannica. As first reported by The Verge, nearly 400 local newspapers recently launched a separate joint action against the same two companies.
The core argument across all these cases is similar: AI chatbots reduce the reason readers have to visit a news site directly, which eats into subscription and advertising revenue that publishers depend on to pay journalists.
| Plaintiff | Filed against | Key demand |
|---|---|---|
| The New York Times | OpenAI, Microsoft | Damages, model changes |
| Ziff Davis | OpenAI | Damages |
| Merriam-Webster / Britannica | OpenAI | Damages |
| ~400 local newspapers | OpenAI, Microsoft | Damages |
| Seattle Times / Newsday | OpenAI, Microsoft | Damages + model destruction |
What does this mean for people who use ChatGPT or Copilot?
For now, nothing changes. Lawsuits take years. No court has yet ordered any AI model destroyed, and OpenAI and Microsoft will almost certainly contest these suits at every stage.
But the legal pressure is real, and it is building. If courts eventually side with publishers on the destruction question, it could force companies to retrain AI systems from scratch using only content they have licensed.
Common questions
Why can't newspapers just block AI companies from using their articles?
Many have tried. Publishers can add instructions to their websites telling web-crawlers not to collect their pages, but critics argue AI companies ignored or worked around those signals when building earlier training datasets. The lawsuits are partly an attempt to enforce consequences after the fact.
Could a court really order an AI model deleted?
It is legally possible but practically complex. Courts can order the destruction of infringing goods, a principle used in physical piracy cases for decades. Whether that doctrine extends cleanly to a software model trained on many sources at once is the live legal question these cases will eventually force judges to answer.
