The judge ruled that prosecutors could compel OpenAI to unlock the ChatGPT account records of Richard Kim, the former chief executive of cryptocurrency startup Zero Edge, as part of a fraud case alleging he diverted investors’ money from crypto trading and online gambling.
Prosecutors allege that Kim diverted approximately $3.8 million from Zero Edge’s $4.3 million fundraising round and that, following his arrest, he spent. ChatGPT researching his case, including his trial strategy. Court filings indicate that Kim may have used AI to carry out numerous orders related to misappropriating investors’ funds, cryptocurrency trading and gambling. Kim has denied security and fraud charges.
US District Judge Lorna Schofield on Monday rejected the defense’s attempt to protect the chatbot’s data, finding that the AI’s chat logs could be considered third-party digital evidence under a search warrant. The warrant seeks Kim’s OpenAI records from October 2023 to May 2026, including notifications, responses and account information.
The case is a reminder that conversations with AI chatbots can become part of the legal record. As more people use tools like ChatGPT for research and personal advice, courts are starting to treat AI conversations like other digital records, like your emails, texts and search history. That also means that using a chatbot for legal research does not automatically make the conversation private or protected.
Can a chatbot be protected by attorney-client privilege?
Kim’s lawyers tried to block the search warrant, saying the chatbot data contained privileged information and research related to the case. According to the defense, those digital records should be protected, as they would reveal Kim’s inner thoughts, defense tactics and prosecution strategies.
Prosecutors argued that, in order to exercise the attorney-client privilege, the conversation must be confidential between an individual and a licensed legal professional for the purpose of obtaining legal advice. An AI chatbot cannot be a lawyer.
Schofield’s decision does not decide whether Kim’s ChatGPT records are protected by the attorney-client privilege, but it does allow the consent to stand, meaning the immunity will not prevent OpenAI from complying. Kim may still challenge some records later.
The dispute adds to a growing legal question about whether instrument negotiations are voluntary ChatGPT, Gemini or Claude they can remain confidential if used for legal research.
The ruling follows the landmark decision earlier this year in United States v. Heppner, where another Manhattan judge found that the defendant’s exchange with Anthropic’s Claude chatbot was not protected by attorney-client privilege or work product protection.
In that case, US District Judge Jed Rakoff said AI platforms are third-party data collectors, not legal advisers. He noted that the defendant used Claude without the guidance of his lawyers and that the forum’s privacy policy undermines any claim that the conversations were private. Rakoff said attorney-directed use of AI can be handled differently.
Future cases may draw sharp lines where AI-assisted legal work can be protected, especially when a chatbot is used at the behest of a lawyer.
It’s not exactly a digital version of Miranda rights, but the caveat here is the same: Anything you write can (and will) be used against you in a court of law.
(Disclosure: Ziff Davis, CNET’s parent company, sued OpenAI in 2025, alleging that it infringed on Ziff Davis’ copyrights in training and using its AI programs.)
