The New York Times, the Daily News and other news publishers have asked a federal judge in Manhattan to sanction OpenAI, alleging that the company obstructed discovery by withholding or destroying evidence relevant to their copyright cases.
A deposition sharpened the dispute
According to an Associated Press report, the publishers’ filing says a recent deposition of an OpenAI employee contradicted earlier representations about the company’s ability to search its training datasets and ChatGPT logs.
The publishers accuse OpenAI of choosing obstruction rather than releasing material that could help determine whether its systems reproduced copyrighted journalism. Those assertions are allegations made in a sanctions motion, not findings by the court.
Steven Lieberman, an attorney representing the Daily News and seven affiliated newspapers, said OpenAI had made misrepresentations for two years about its ability to search for copyrighted material in its datasets and logs.
What the publishers are seeking
The news organisations want the judge to penalise OpenAI for what they describe as discovery misconduct. The requested relief includes reimbursement of legal fees spent pursuing evidence that the publishers say was improperly withheld.
The sanctions dispute concerns how evidence was preserved and produced during litigation. It is separate from the underlying question of whether OpenAI infringed copyright when developing and operating its models.
OpenAI denies the allegations
OpenAI has rejected the publishers’ account. Spokesperson Drew Pusateri called the allegations “blatantly false” and said The New York Times was attempting to invade the privacy of users who have no connection to the litigation.
The company said it would continue defending its users’ privacy and what it described as the long-established principles of fair use. OpenAI has previously argued that restrictions on sharing ChatGPT logs are necessary because the records can contain sensitive user information.
The broader copyright fight
The New York Times sued OpenAI and Microsoft in late 2023, alleging that its journalism had been copied and used to develop AI products that can compete with the original reporting. Other news organisations have since brought or joined related cases.
OpenAI and other technology companies maintain that training AI systems on material found online can qualify as fair use. Publishers argue that the systems were built using costly original reporting without permission or compensation and can produce substitutes that reduce traffic to the original sources.
What comes next
The court must now decide whether OpenAI’s conduct during discovery warrants sanctions and, if so, what remedies would be proportionate. Any decision could influence how AI companies are expected to preserve, search and produce internal datasets and user-output records in future copyright cases.
The sanctions ruling will not by itself decide whether OpenAI infringed the publishers’ copyrights. But it could determine what evidence the parties may rely on when that larger question eventually reaches the court.
