OpenAI fights court order to retain user ChatGPT logs in NYT battle

When a copyright battle drags on long enough, ordinary user logs often become the surprise collateral damage.

OpenAI officially pushed back against a court preservation order stemming from its ongoing copyright lawsuit with The New York Times. The news plaintiffs had demanded that OpenAI indefinitely preserve all consumer ChatGPT and API logs—including conversations that users deliberately deleted.

OpenAI called the demand an unacceptable overreach. Chief Operating Officer Brad Lightcap warned that freezing user deletions directly violates established privacy norms, vowing that the company would fight the order in court.

Deleted chats caught in legal crossfire

The dispute centers on an order issued on May 13 by Magistrate Judge Ona Wang. News plaintiffs argued that web users might use ChatGPT to bypass newspaper paywalls, and that those readers would likely delete their search logs to cover their tracks.

Labeling OpenAI’s routine 30-day deletion protocol as evidence destruction, the plaintiffs convinced the court to order OpenAI to retain and segregate all output log data that would otherwise be permanently scrubbed. Under standard policies, deleted conversations in ChatGPT Free, Plus, Pro, and Team, along with standard API inputs and outputs, are purged within 30 days.

The sweeping preservation order threatened the privacy of hundreds of millions of users worldwide. Only ChatGPT Enterprise and API clients holding signed Zero Data Retention (ZDR) agreements remained unaffected, as the court later clarified and ZDR setups never store logs to begin with.

OpenAI pushed back hard, appealing the order while locking down the affected data. The company confirmed that the retained logs were kept in a separate secure vault, strictly inaccessible to the plaintiffs or the court, and monitored only by a tiny, audited legal and security team.

By October 22, 2025, OpenAI updated its notice to announce a major relief: the mandate to indefinitely retain ongoing user data expired on September 26, 2025. Standard 30-day automatic deletion has since resumed across the board.

However, The New York Times continues to demand historical user logs spanning April through September 2025. While OpenAI must keep that specific window locked down for now, it is refusing to hand over the files and continues to challenge the demand.

Why it matters

This legal showdown is a loud wake-up call for anyone treating an AI chat prompt like a private journal.

For regular users and independent developers, the immediate takeaway is clear: your delete button is only as solid as the legal jurisdiction behind it. Even if you wipe sensitive code or personal thoughts, a sweeping court order in an unrelated lawsuit can freeze your logs in a server vault. If you are handling truly confidential work, moving to ChatGPT Enterprise or a dedicated Zero Data Retention API is no longer an optional upgrade—it is a baseline requirement.

For the AI sector, the case reveals an alarming shift in litigation strategy. Copyright battles used to revolve strictly around pre-training datasets; now, plaintiffs are probing real-time inference logs, arguing that ordinary users might be infringing content on the fly. If that logic sticks, AI providers could routinely be forced into surveillance roles under the guise of evidence preservation.

OpenAI handled this tactically well. By framing the battle as a principled stand for user privacy—even citing The New York Times' own 2020 editorial advocating that people control their personal data—it effectively turned a messy copyright issue into a high-ground defense of consumer trust. Still, keeping months of locked historical logs unresolved means the battle over who truly controls conversational data remains wide open.