Judge orders OpenAI to preserve all ChatGPT logs in NYT copyright case
Magistrate Judge Ona Wang ordered indefinite retention of chats users had deleted, calling roughly 20 million logs relevant to the New York Times' copyright claims.
- Courts & copyright
- Notable
In the discovery phase of The New York Times’ copyright suit against OpenAI, Magistrate Judge Ona Wang of the Southern District of New York ordered OpenAI to “preserve and segregate all output log data that would otherwise have been deleted,” effective immediately and until further order of the court. The order covered chats from ChatGPT Free, Plus, Pro and Team users, and API users without a zero-data-retention agreement, overriding OpenAI’s normal policy of permanently deleting conversations users chose to remove or that expired after 30 days as temporary chats.
The Times had asked for the order after learning OpenAI was continuing to delete logs following the lawsuit’s filing, arguing a statistically valid sample of conversations was needed as evidence that ChatGPT could reproduce or closely paraphrase its articles. The judge agreed the logs were relevant and, weighing OpenAI’s objections over cost and user privacy against the risk of lost evidence, sided with preservation. The practical scope settled at roughly 20 million anonymised logs, a small fraction of OpenAI’s total conversation volume but far more than the company wanted to hand over.
OpenAI objected publicly, calling the order a privacy “overreach” that abandoned the company’s normal data-retention commitments to users, and sought reconsideration; the motion was denied on 16 May. The company argued the requirement conflicted with privacy laws in Europe and elsewhere that give users a right to have their data deleted, and said it would have preferred to fight the demand rather than comply. Privacy advocates and some security researchers raised a related concern: a large, centralised store of ordinarily-deleted conversations, however anonymised, was itself a new target, and its existence was dictated by litigation strategy rather than by any operational or safety need.
The order remained in effect for months — a burden OpenAI continued to contest publicly even as compliance proceeded — before the same judge narrowed it that October, allowing the company to resume its normal deletion policy for most users going forward while preserving what had already been retained.