Two more regional papers sued OpenAI and Microsoft. Same week Google started paying publishers.

What happened

The Seattle Times and Newsday have filed copyright claims against OpenAI and Microsoft over the use of their journalism in AI models, according to Law Commentary's report on the filing. Two regional papers, not a national chain. That is the detail worth noticing.

The claims track the shape of the New York Times case: published work ingested, models trained, output that competes with the original. The filings do not open a new legal theory. They widen the plaintiff pool.

What is not confirmed from the reporting: damages sought, the exact scope of the works at issue, and whether either paper had prior licensing talks with either company. Do not fill those blanks in your head.

Why this lands differently than the NYT case

The New York Times has a licensing business, a legal department, and leverage. The Seattle Times and Newsday are the tier below — regional operations with deep archives and no seat at the table when platforms hand out deals.

When that tier starts filing, it stops being a story about a few large rights holders and starts being a question about whether any archive has a price.

The legal weather is not friendly right now. The DOJ told a court it should rule for OpenAI in the Times case. And in the George R.R. Martin matter, a judge took up summary judgment this month, per Books of Brilliance — the stage where a case either survives to trial or ends. Courts are starting to answer questions that had been open for two years.

The two tracks, running at once

Same week, opposite direction: Digiday reported on Sept. 14 that Google is rolling out a pay-per-value program that pays publishers when their content significantly contributes to answers in Gemini, AI Overviews and AI Mode.

So there are now two prices on your work, and they are not the same price.

One is training — the one-time ingestion of your archive. That is what the lawsuits are about, and it is being settled in courtrooms, slowly, with outcomes you do not control.

The other is use in answers — your work cited or drawn on in a live response. That is what Google says it will pay for, through a dashboard, on terms nobody has published.

A publisher can lose the first fight entirely and still collect on the second. They are separate revenue events with separate mechanics.

What this is not

It is not a signal that suing is now the winning play. Two filings is not a trend, and neither case has produced a ruling.

It is not a reason to hold out on licensing offers. Nothing in the reporting suggests these papers turned down money. The opposite reading — that regional publishers sue because nobody offered — is at least as plausible and equally unconfirmed.

And it does not touch video or audio. These are text archive claims. If your body of work is on YouTube or in a podcast feed, this case does not cover you.

What to do about it

1. Separate your archive into two buckets in writing: material a model may have trained on, and material that gets cited in live AI answers today. They negotiate separately.
2. Check whether you actually hold the rights to your back catalog. Work-for-hire, co-host agreements and old contributor deals decide whether you have standing at all.
3. If you have an AI licensing offer on the table, do not pause it waiting for a court. No ruling in these cases lands this quarter.
4. Log your AI Overviews impressions-versus-clicks now. Whatever any deal pays later, this is the baseline you judge it against.
5. Set a 60-day check on the Martin summary judgment. A ruling there moves the price of every archive more than a new filing does.