FULL STORY

NYT vs. OpenAI: Court Orders Chat Log Disclosure

The NYT accused OpenAI of withholding evidence regarding its training data. The court subsequently ordered OpenAI to preserve and disclose millions of anonymized ChatGPT conversation logs to the plaintiffs.

2026-07-10 ~ 2026-07-13 · 3 episodes · 18 posts

Episode 1 · NYT and publishers seek sanctions over alleged OpenAI evidence concealment (2026-07-10, 12 posts)

The copyright fight around ChatGPT has escalated from a dispute over training data and outputs into a procedural battle over evidence. The New York Times and other plaintiffs are asking a federal court to sanction OpenAI, arguing that the company concealed or misrepresented its ability to search training data and chat logs—an issue they say is central to proving whether ChatGPT reproduced copyrighted news content.

Key details

According to the posts, 17 publishers joined the push for sanctions. The plaintiffs say later testimony, including April testimony from an OpenAI engineer, showed OpenAI had capabilities it had not properly disclosed in court, including searching data and maintaining a de-identified chat-log database. One cited figure is 78 million chat records that, the plaintiffs argue, had begun to accumulate before The New York Times filed suit.

The Times has sought chat logs to help show that ChatGPT could reproduce its copyrighted reporting, whether from training data or web-connected retrieval. OpenAI had argued in court that producing such material would harm user privacy and that de-identifying the data would be burdensome. The plaintiffs say the later testimony undercuts those representations.

Sanctions request and OpenAI response

The sanctions request reportedly includes monetary sanctions and a request to bar OpenAI from relying on a 20 million-record sample in its defense. OpenAI had tried to submit that de-identified sample to the court, but one post says the court found it so heavily redacted that it was nearly unusable.

OpenAI denies the allegations. In the reporting cited in the posts, the company calls the claims "obviously false" and argues that The New York Times is trying to obtain private user data.

Why it matters

This dispute goes beyond whether copyrighted material appeared in model outputs. It now also tests how much visibility courts can demand into AI companies’ internal data practices, and whether privacy and technical-burden arguments hold up when later testimony suggests broader logging or search capabilities than previously described.

Episode 2 · Court Orders OpenAI to Disclose 20M Anonymized Chat Logs (2026-07-10, 4 posts)

In an ongoing copyright infringement lawsuit, a federal court ordered OpenAI to preserve all ChatGPT logs and provide plaintiffs with around 20 million anonymized conversations for keyword searching. OpenAI's objections were overruled, escalating privacy concerns.

Episode 3 · NYT Accuses OpenAI of Withholding Evidence in Copyright Lawsuit (2026-07-13, 2 posts)

The New York Times and other publishers have accused OpenAI of withholding evidence, demanding court sanctions after OpenAI was found to have searched its training data despite claiming otherwise. This copyright dispute is escalating into a fierce legal battle over discovery procedures.