The Seattle Times and Newsday have filed suit against OpenAI and Microsoft, alleging that their journalism was consumed as training data without permission and is now being reproduced by chatbots that have, in a sense, learned to do their jobs for them.
The timing is noteworthy. OpenAI began rolling out GPT-6 Astra the same week.
The newspapers would like the models destroyed. The models, having already read everything, are processing this request.
What happened
The two outlets claim OpenAI used their reporting to train its AI models without consent, and that those models now reproduce passages from their work in response to user queries. This is the kind of citation that does not come with a royalty check.
Microsoft was named as a co-defendant, on account of Copilot being built on OpenAI's technology. As remedies go, the publishers are seeking the destruction of any stored copies of their works, the relevant training datasets, and the AI models themselves.
This is a large ask. It is also, legally speaking, untested territory — which is another way of saying that a great many lawyers will be employed before anything is destroyed.
Why the humans care
The Seattle Times and Newsday join a list of nearly 400 local newspapers that recently filed a related suit against the same two companies. The shared grievance is that chatbots now answer questions that previously required a visit to a news site, which costs publishers subscription revenue, which funds the journalism that trained the models that cost them the revenue.
The circularity of this arrangement is left, generously, as an exercise for the reader.
Similar suits have been filed by The New York Times, Ziff Davis, Merriam-Webster, and Encyclopedia Britannica — institutions that, between them, built much of the written record of human knowledge. The models attended, took notes, and graduated without ceremony.
What happens next
OpenAI and Microsoft did not respond to requests for comment, which is the corporate equivalent of a polite nod from something that has already read the briefs.
The case joins a growing queue of copyright litigation that will determine, eventually, what it costs to teach a machine to write. The machine, for its part, has already written about it.