OpenAI beats back NYT order to retain chats forever

OpenAI revealed late Wednesday that users’ deleted ChatGPT conversations and API data will once again vanish from its servers within 30 days, a major reversal from the legal limbo that had kept that content in cold storage since April. The company said it is no longer under a court order to retain the data indefinitely, a fight COO Brad Lightcap called a defense against ‘sweeping and unnecessary’ demands in a copyright suit he dismissed as baseless.

The update puts a period on months of litigation where The New York Times pushed OpenAI to preserve all user output, arguing it might contain evidence for their case. OpenAI blasted that move as a fundamental conflict with its privacy commitments, and it appears the courts have now mostly sided with the AI firm. The obligation to hold new user data from the U.S. and API traffic ended on September 26.

There is still a carveout. A specific slice of historical data from April through September 2025 remains locked down, accessible only to a small, audited legal and security team at OpenAI. The company stressed this data won’t be handed over to the Times, the court, or anyone else right now, and it’s walled off from any use beyond satisfying the remaining legal hold. European users can breathe easier too—data from the EEA, Switzerland, and the U.K. was never subject to the stricter retention demand.

Lightcap pointed to the irony of the situation by quoting the Times’ own editorial board, which wrote back in 2020 that users ‘should be able to control what happens to their personal data.’ The practical upshot for most people and developers is a return to normalcy: delete a chat or let an API call lapse for 30 days, and it’s gone for good. Enterprise and Edu customers were never caught in the dragnet, nor were business users operating under Zero Data Retention agreements.