A coalition of 16 news publishers alleges OpenAI concealed search capabilities and may have deleted 20 million conversation logs critical to a landmark copyright trial.
A group of roughly 16 news organizations, with The New York Times at the helm, filed a motion on July 9 seeking sanctions against OpenAI in US District Court for the Southern District of New York. The publishers claim the ChatGPT maker hid evidence, misrepresented its technical capabilities, and may have deleted or failed to preserve around 20 million conversation logs that could prove copyright infringement.
What the publishers are alleging #
The coalition, which includes the New York Daily News and Chicago Tribune alongside the Times, accuses OpenAI of concealing its ability to search training datasets for over two years. When asked whether it could identify copyrighted materials used to train its models, OpenAI allegedly said it couldn’t. The publishers now claim that was false.
The missing conversation logs are central to the dispute. Those 20 million ChatGPT output logs, the publishers argue, could demonstrate that OpenAI’s models reproduced or closely mimicked copyrighted news articles. Without them, the plaintiffs say they’re being denied evidence essential to their case.
The sanctions motion asks for monetary penalties, including attorneys’ fees, and adverse findings. In plain English: the publishers want the court to assume the missing logs would have supported their infringement claims. That kind of ruling could be devastating for OpenAI’s defense.
Notably, Microsoft is not a target of this particular sanctions filing, even though it was named as a co-defendant in the original lawsuit. The underlying case accuses both OpenAI and Microsoft of using millions of news articles to train large language models like ChatGPT without permission or compensation.
The bigger legal picture #
This case has been building for a while. The original copyright infringement lawsuit was filed in December 2023, making it one of the earliest and highest-profile legal challenges to how AI companies source their training data.
Both OpenAI and Microsoft filed motions to dismiss in 2025. The court denied those motions on the core copyright claims, which survived and are now in the discovery phase. That’s the stage where both sides exchange evidence, and it’s precisely where the publishers say OpenAI is stonewalling.
The sanctions motion represents a significant escalation. Courts don’t take evidence destruction lightly, and allegations that a defendant deliberately deleted relevant records or lied about its technical capabilities can reshape a trial’s trajectory before it even reaches a jury.
For context, sanctions in federal litigation aren’t just about money. If a judge agrees that OpenAI failed to preserve evidence, the court can instruct the jury to assume that the destroyed evidence was unfavorable to OpenAI. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our