A U.S. federal court has ordered OpenAI to indefinitely preserve all user-generated content from ChatGPT and its API services, including chats that users have deleted.
The preservation order, signed by Magistrate Judge Ona T. Wang of the Southern District of New York, stems from ongoing copyright litigation initiated by The New York Times and other media organizations.
The court’s directive compels OpenAI to retain all “output log data that would otherwise be deleted,” citing concerns that relevant evidence could be lost. Plaintiffs argue that deleted user interactions may reveal attempts to generate paywalled or copyrighted news content using ChatGPT, and allege that OpenAI’s default deletion policies may have already led to the destruction of key data.
The litigation is part of a broader multi-district case (MDL No. 25-md-3143), wherein The New York Times and other outlets allege copyright infringement by OpenAI. Their renewed motion, prompted by suspicions that users might intentionally erase evidence of misuse, contended that OpenAI’s deletion protocols impeded their ability to prove infringement. Judge Wang noted that OpenAI had not committed to preserving or segregating deletion-marked data unless legally compelled, prompting the issuance of the order.
OpenAI, the maker of the ChatGPT chatbot and a major player in generative AI, responded sharply. Chief Operating Officer Brad Lightcap decried the court’s mandate as “sweeping and unnecessary,” arguing that it clashes with OpenAI’s longstanding privacy policies and undermines user trust. In a public statement, the company warned that the order jeopardizes the privacy of hundreds of millions of users globally and could force it into conflict with both contractual commitments and international privacy laws, including the EU’s GDPR.
“This fundamentally conflicts with the privacy commitments we have made to our users,” Lightcap said. “It abandons long-standing privacy norms and weakens privacy protections.” OpenAI is appealing the order and has requested oral arguments in hopes of reversing or narrowing the mandate.
The scope of the court’s order is expansive. It affects all users of ChatGPT Free, Plus, Pro, and Team, as well as developers using OpenAI’s API, unless they are covered by a Zero Data Retention (ZDR) agreement. Enterprise and education customers are exempt, as are any API users employing ZDR endpoints, which OpenAI says do not log any data by design.
Until the order is lifted, OpenAI is obligated to store even chats that were manually deleted or entered in “Temporary Chat” sessions that would normally vanish after closure. The company clarified that retained data is being stored in a secure system under legal hold, accessible only to a small, audited legal and security team and not automatically shared with plaintiffs.
Privacy advocates and users have expressed outrage, flooding social media with concerns about the implications of the order. Many pointed out that ChatGPT sessions often contain sensitive personal data, including financial records, medical information, and business trade secrets. Critics argue the order could deter users from engaging freely with AI tools and prompt a shift toward rival platforms that offer stronger data deletion assurances.
In its appeal, OpenAI insists that there’s no concrete evidence of data destruction in response to the lawsuit and that the court’s decision was based on speculative harm. The company also emphasized the engineering burden required to comply with the order, warning that it diverts substantial resources away from product development and user safety.
Leave a Reply