About

Federal court says public AI chats can waive privilege in *Heppner* ruling

Published
Score
11

Why it matters

A federal judge in the Southern District of New York has ruled that a criminal defendant's conversations with a publicly available generative AI platform fall outside attorney-client privilege and the work-product doctrine, potentially exposing them to prosecution. In United States v. Bradley Heppner, Judge Jed S. Rakoff held that exchanges with consumer AI tools like Claude are not protected communications simply because they are later shared with counsel. The court reasoned that communications with public, non-enterprise AI systems lack the confidentiality required for privilege protection and do not become privileged retroactively through disclosure to an attorney.

The decision marks the first federal ruling squarely addressing whether client communications with consumer AI platforms qualify for privilege. The specific scope of Heppner's AI exchanges and the precise nature of the materials prosecutors sought to use remain unclear from available filings.

Attorneys should treat this ruling as a concrete warning about discovery and ethics risks. As lawyers and clients increasingly use ChatGPT, Claude, and similar tools for sensitive work, the Heppner decision establishes that public AI systems create waiver exposure. Because these platforms may retain, store, or train on user inputs, entering confidential information—case strategy, client facts, investigative findings—into a consumer AI tool can destroy privilege protection. The ruling has immediate implications for criminal defense, civil discovery, corporate compliance, and attorney ethics guidance. Law firms are already updating policies on AI governance and vendor controls, but Heppner signals that the safest approach remains: do not input privileged or confidential information into public AI systems at all.

mail Subscribe to Law And Technology email updates

Primary sources. No fluff. Straight to your inbox.

Also on LawSnap