Immediate production of the output log sample is essential to stay on track for the February 26, 2026, discovery deadline. OpenAI’s proposal to run searches on this small subset of its model outputs on Plaintiffs’ behalf is as inefficient as it is inadequate to allow Plaintiffs to fairly analyze how “real world” users interact with a core product at the center of this litigation. Plaintiffs cannot reasonably conduct expert analyses about how OpenAI’s models function in its core consumer-facing product, how retrieval augmented generation (“RAG”) functions to deliver news content, how consumers interact with that product, and the frequency of hallucinations without access to the model outputs themselves.

OpenAI said the NYT’s discovery requests were initially limited to logs “related to Times content” and that it has “been working to satisfy those requests by sampling conversation logs. Towards the end of that process, News Plaintiffs filed a motion with a new demand: that instead of finding and producing logs that are ‘related to Times content,’ OpenAI should hand over the entire 20 million-log sample ‘via hard drive.’”

OpenAI disputes judge’s reasoning

The November 7 order cited a California case, Concord Music Group, Inc. v. Anthropic PBC, in which US District Magistrate Judge Susan van Keulen ordered the production of 5 million records. OpenAI consistently relied on van Keulen’s use of a sample-size formula “in support of its previous proposed methodology for conversation data sampling, but fails to explain why Judge [van] Keulen’s subsequent order directing production of the entire 5 million-record sample to the plaintiff in that case is not similarly instructive here,” Wang wrote.

OpenAI’s filing today said the company was never given an opportunity to explain why Concord shouldn’t apply in this case because the news plaintiffs did not reference it in their motion.

“The cited Concord order was not about whether wholesale production of the sample was appropriate; it was about the mechanism through which Anthropic would effectuate an already agreed-upon production,” OpenAI wrote. “Nothing about that order suggests that Judge van Keulen would have ordered wholesale production had Anthropic raised the privacy concerns that OpenAI has raised throughout this case.”

The Concord logs were just prompt-output pairs, “i.e., a single user prompt followed by a single model output,” OpenAI wrote. “The logs at issue here are complete conversations: each log in the 20 million sample represents a complete exchange of multiple prompt-output pairs between a user and ChatGPT.” That could result in “up to 80 million prompt-output pairs,” OpenAI said.