“A snake eating its own tail.” That’s how The Seattle Times and Newsday describe generative AI in their newly filed lawsuit against OpenAI and Microsoft, and it’s probably the most accurate framing of this dispute yet. The two publishers filed the lawsuit on Friday, accusing both companies of training AI models on their copyrighted journalism without permission.
The core allegations are specific. Seattle Times and Newsday claim OpenAI and Microsoft were methodically scraping news articles in ways that bypass paywalls, essentially treating paid content as free training material. The publishers argue this has directly damaged their business, because users can now get AI-generated summaries instead of visiting the actual articles. Less traffic means less ad revenue. And when AI models hallucinate and attribute false information to these outlets, the reputational damage compounds the financial one. The lawsuit also flags the removal of copyright management information from articles, which adds another layer to the legal argument.
This is not an isolated case. The New York Times set the precedent in 2023 when it sued OpenAI and Microsoft on similar grounds. That case is still working through the courts, and it has clearly encouraged other publishers to pursue the same route. CNN sued Perplexity for copyright infringement earlier this year. The pattern is consistent: news organizations are watching AI companies profit from content they spent money and time to produce, without compensation or even formal acknowledgment.
What makes this moment interesting is the split forming within the media industry. Not everyone is suing. The Associated Press and Vox Media have both struck licensing deals with OpenAI, choosing partnership over litigation. That divide matters, because it signals there’s no unified industry position. Some publishers believe they can negotiate their way into the AI economy. Others think the only way to establish real rights is through the courts.
For developers and founders building on top of OpenAI or Microsoft’s models, this legal pressure is worth tracking closely. If courts start ruling against AI companies on training data, it could force changes to how models are built, what data is permissible, and whether licensing costs get passed downstream. The New York Times case alone is already being watched as a potential bellwether. Adding more plaintiffs and more jurisdictions only increases the pressure on OpenAI and Microsoft to either settle or fight this out publicly. Neither option is clean.
