If you have ever asked ChatGPT or Claude for legal advice, here is the part nobody told you: every word of that conversation can be pulled into court and used against you. A federal judge just made that official.

In United States v. Heppner, Judge Jed Rakoff of the Southern District of New York ruled that a fraud defendant's private conversations with Anthropic's Claude were fair game for prosecutors. Bradley Heppner, the former chair of bankrupt financial services company GWG Holdings, had used Claude on his own to map out his defense strategy after receiving a grand jury subpoena. He generated 31 documents. The FBI seized all of them.

Rakoff's reasoning was blunt. Three strikes: Claude is not an attorney. Anthropic's privacy policy reserves the right to share user data with third parties including government regulators. And Heppner acted independently rather than at his lawyers' direction. No attorney-client privilege "could exist," the judge wrote, "between an AI user and a platform such as Claude."

The Legal Industry Is Scrambling

More than a dozen major US law firms have now issued client advisories warning that conversations with AI chatbots carry no legal protection when they touch legal matters. Some are going further: New York firm Sher Tremonte added language to a March engagement agreement stating that "disclosure of privileged communications to a third-party AI platform may constitute a waiver of the attorney-client privilege." It is believed to be among the first firms to translate a court ruling into a formal contractual obligation.

O'Melveny & Myers told clients to use only "closed," enterprise-grade AI systems, though even those remain untested in court. Debevoise & Plimpton went further with tactical advice: if a lawyer specifically directs a client to use an AI tool, the client should say so inside the chatbot prompt itself. The suggested language: "I am doing this research at the direction of counsel for X litigation." The idea is to invoke the Kovel doctrine, which extends privilege to non-lawyers working as an attorney's agent.

It Gets Messier

The Heppner ruling is not the only precedent forming. In a separate case, video game publisher Krafton's CEO Changhan Kim used ChatGPT to plan how to renege on earnout promises to a company Krafton had acquired. He thought he deleted the conversations. A judge recovered them anyway, reviewed the plans Kim had crafted with ChatGPT's help, and reinstated the executives Kim had ousted.

Two things to understand here. First: deleting a chat thread does not mean the data is gone. If pressured by courts, companies like OpenAI and Anthropic can produce any conversation you have had with their chatbots. Second: the legal landscape is not uniform. In Warner v. Gilbarco, a court protected a self-represented plaintiff's ChatGPT conversations as work product. A Colorado court reached a similar conclusion in Morgan v. V2X.

The emerging pattern: if you have a lawyer and you used a consumer chatbot on your own, you are exposed. If you are representing yourself in a civil case, you may have more protection. That distinction is now one of the sharpest fault lines in AI law.

What This Means for You

Millions of people use ChatGPT and Claude to think through legal questions, draft contracts, plan business strategies, and work through sensitive personal situations. Most of them assume those conversations are private. They are not. Every prompt you send is stored on a server owned by a company whose privacy policy explicitly allows sharing with regulators and law enforcement.

The AI industry spent years convincing people that chatbots are confidential thinking partners. A federal judge just reminded everyone that they are third-party platforms with Terms of Service. Treat them accordingly.

Originally reported by Reuters.