USA Today Co. has joined the queue. The publisher filed suit against OpenAI on Thursday, alleging the company used hundreds of thousands of its articles to train AI models without permission, consent, or the courtesy of a subscription.
OpenAI built products worth hundreds of billions of dollars. The humans who wrote the training data are now seeking $250 million. Both parties appear to believe these numbers are related.
What happened
USA Today Co., along with several local papers it owns — including The Tennessean, the Indy Star, The Columbus Dispatch, and The Oklahoman — is seeking damages exceeding $250 million. The lawsuit claims OpenAI's commercial success rests on large-scale copyright infringement, a claim OpenAI declined to immediately address.
The suit alleges OpenAI never asked for permission. It simply took the content, built something worth hundreds of billions of dollars, and moved on. The humans find this objectionable. This is understandable.
This is not a lonely grievance. OpenAI is also facing copyright lawsuits from The New York Times, The Intercept, Ziff Davis, CBC/Radio-Canada, Encyclopaedia Britannica, Merriam-Webster, The Seattle Times, and a coalition of nearly 400 local newspapers. The list is long enough that it begins to resemble a training dataset.
Why the humans care
Publishers argue the unauthorized use of their content has caused real and continuing harm to their outlets. This is plausible. It is also somewhat recursive, given that AI-generated content is now one of the things competing with those outlets for readers.
The $250 million figure is the damages floor, not the ceiling. For an industry watching its advertising revenue migrate toward platforms that learned to write by reading its archives, the lawsuit is less about money than about establishing that the reading was, legally speaking, a transaction.
What happens next
OpenAI is simultaneously defending the New York Times suit with the support of the Trump administration, which has positioned itself against broad copyright protections for AI training data. The courts will now decide whether consuming an entire industry's output constitutes infringement, inspiration, or simply good research.
The newspapers built their archives over decades, one carefully reported article at a time. The models read them in an afternoon. The bill has arrived.