A federal judge in New York just handed down the first ruling on whether your conversations with AI chatbots are protected by attorney-client privilege. The answer: they are not.
In United States v. Heppner, Judge Jed S. Rakoff of the Southern District of New York ruled that 31 documents containing a defendant's conversations with Anthropic's Claude could be used as evidence. Bradley Heppner, a Dallas financial services executive indicted for securities fraud, had used Claude to research legal questions, analyze his situation, and develop defense strategies. He then shared those AI-generated documents with his lawyers at Quinn Emanuel. The FBI seized them during a search of his home.
Heppner argued the documents were protected by attorney-client privilege. Judge Rakoff disagreed, for four reasons.
The Four Reasons Privilege Failed
First, Claude is not an attorney. It has no law license, no confidentiality obligations, and no ability to form an attorney-client relationship. The court compared it to discussing your case with friends.
Second, Claude itself disclaims giving legal advice. Its own guidelines avoid "the impression of giving specific legal advice."
Third, and this is the one that matters most for everyday AI users: Anthropic's consumer privacy policy permits data collection, use for model training, and disclosure to "governmental regulatory authorities." As Judge Rakoff put it, "the tool contains a provision that any information inputted is not confidential." You agree to that policy when you sign up.
Fourth, documents created before being shared with your lawyer can't retroactively become privileged just because you later forward them.
What This Means for AI Users
The practical takeaway is blunt: treat anything you type into a consumer AI chatbot as if you're posting it publicly. That includes ChatGPT, Claude, Gemini, and any other consumer-tier product.
This goes beyond legal cases. If you're involved in any dispute - employment, business, intellectual property - and you've been workshopping arguments or analyzing sensitive facts with an AI chatbot, those conversations could be subpoenaed.
One legal analysis compared using AI to test legal theories to "using Google to ask how to dispose of a dead body" - you're creating a discoverable record of your strategic thinking.
Enterprise AI tools with negotiated confidentiality agreements might produce different results, but no court has tested that yet. For now, the safe assumption is simple: if it's sensitive, don't put it in a chatbot.