• by Ray Schultz , Columnist, July 16, 2025

A federal judge has had it with the bickering over discovery in the copyright case filed by the New York Times and other publishers against OpenAI.  

U.S. Magistrate Judge Ona T. Wang has given the parties until this Friday to file a “single joint proposed interim Protective Order” covering documents that OpenAI wants excluded.  

This would be “applicable to all cases, incorporating the Court’s rulings thus far,” Wang rules. 

A hint of irritation is apparent in some of the judge’s language.  

“It is not clear, nor does OpenAI provide any reason, why the parties should not work together to identify and segregate the documents that fall into any of the seven Excluded Documents categories and why engaging in this identification process would not be proportional to the needs of the case,” Wang writes in the ruling issued on Wednesday.

In essence, Wang seems to be telling OpenAI to stop whining and get on it. 

 “For example, it seems improbable that OpenAI would have reviewed, designated and produced approximately 13,500 AEO (attorneys’ eyes only)documents without also categorizing (or developing tools to find and categorize) them,” the judge notes. “Similarly, News Plaintiffs likely have tools to conduct searches that could categorize and then segregate Excluded Documents, although OpenAI is more likely to have a better understanding of its own documents.”

Want directs the sides to file a joint status letter updating the court on the status of the procedural dispute by July 23. 

However, Wang writes that the court “adopts OpenAI’s proposal to exclude documents that reveal ‘strategies for development of customer or partnership relationships’ based on OpenAI’s representation to the Court that such language ‘would not apply to documents ‘addressing how users interact with ChatGPT.’ The parties are directed to include this representation by OpenAI in the proposed Amended Protective Order.”

To repeat what has often been said, the case is about the defendants’ “misappropriation of millions of News Plaintiffs’ articles for training their large language models and building generative AI products that can deliver those articles, or derivatives thereof, to users as output,” the plaintiffs contend.  

If nothing else, this shows that publishers eager to take on companies like OpenAI better have strong legal teams, deep pockets and patience. 

The case is on file with the U.S. District Court or the Southern District of New York.