Federal Judge Rules AI Chat Communications Lack Legal Privilege
A defendant's communications with an AI platform cannot be protected by attorney-client privilege or work product doctrine, a federal judge ruled in April 2026. The decision in U.S. v. Heppner marks the first nationwide ruling on whether conversations with publicly available AI tools receive legal protection in criminal investigations.
Bradley Heppner received a grand jury subpoena in late 2025 targeting him in a securities and wire fraud investigation. Without consulting his lawyer, Heppner used Claude to draft defense strategy documents and arguments based on anticipated charges. The FBI later seized 31 documents from that exchange under a valid search warrant.
Heppner argued the communications should be privileged because he was preparing for potential indictment and had shared the documents with counsel. Judge Jed S. Rakoff of the Southern District of New York rejected both arguments.
Why Attorney-Client Privilege Failed
The judge found three fatal flaws in Heppner's privilege claim. First, Claude is not an attorney, so no attorney-client relationship existed.
Second, Heppner could not reasonably expect confidentiality. Claude's user policy-which Heppner agreed to-states that Anthropic collects data on user inputs and outputs and reserves the right to share it with third parties. The policy explicitly allows disclosure "in connection with claims, disputes, or litigation."
Third, Heppner did not communicate with Claude to obtain legal advice. Claude itself tells users it cannot provide formal legal advice, and Heppner's counsel never directed him to use the platform.
Work Product Doctrine Also Fell Short
The work product doctrine protects materials prepared by or at the direction of counsel in anticipation of litigation. Heppner prepared the documents alone, without instruction from his attorney. His counsel acknowledged the materials did not reflect their litigation strategy at the time Heppner created them.
Judge Rakoff said the case presented a straightforward application of existing law. "AI's novelty does not mean that its use is not subject to longstanding legal principles," he wrote.
What This Means for Communications Professionals
The ruling has immediate practical implications for anyone in PR or communications roles who work with legal teams or handle sensitive matters. Using AI platforms to draft strategy, analyze risks, or prepare responses to investigations creates discoverable records without privilege protection.
Communications professionals should treat AI-generated documents the same way they would treat unprotected memos. If legal advice is needed, direct consultation with counsel remains essential. If you use AI tools, assume those conversations are not confidential and could be disclosed in litigation or investigations.
The decision also signals how courts will approach AI and privilege questions. Judges will focus on whether the AI provider can access user data and whether the user could reasonably expect privacy-not on how the AI tool was used or what work it performed.
Learn more about Claude AI and its capabilities, or explore AI for Legal Professionals Courses to understand how AI intersects with legal compliance and privilege in your organization.
Your membership also unlocks:
