USDC SDNY: Client Generated AI Documents Not Protected by Attorney-Client Privilege

USDC SDNY: Client Generated AI Documents Not Protected by Attorney-Client Privilege

On February 17, 2026, the U.S. District Court of the Southern District of New York found in a first-of-its-kind ruling that written exchanges between a criminal defendant and a generative artificial intelligence (“AI”) platform were not protected from disclosure by attorney-client privilege or attorney work product doctrine. The ruling in United States v. Heppner should give pause to clients who are inclined to turn to AI products to help summarize or understand their legal claims both before and after speaking with attorneys.

What happened: United States v. Heppner, Case 1:25-cr-00503-JSR (Dkt. 27, Feb. 17, 2026)

Defendant Bradley Heppner was indicted by a grand jury on various counts related to securities fraud on October 28, 2025. When the Federal Bureau of Investigation arrested Heppner on November 4, 2025, they also executed a search warrant at his home where they seized documents and electronic devices. Among the documents seized included approximately 30 documents memorializing interactions between Heppner and the Anthropic AI platform known as “Claude.” According to Heppner’s counsel, the communications with Claude were generated after Heppner became aware that he was a target of the government investigation that led to his arrest, and after he had received a grand jury subpoena on the matter. Heppner’s counsel argued that the communications with Claude should be considered attorney-client privileged because they were generated in anticipation of a potential indictment, included information that Heppner had learned from communications with his attorney, and were prepared by Heppner for the purpose of speaking with counsel to obtain legal advice. The Court dismissed all three arguments in turn. The Court held that the assertion of the attorney-client privilege requires a trusting human relationship with a licensed professional subject to ordinary fiduciary duties and discipline procedures. It further held that the use of third-party service like Claude with an explicit privacy policy that reserves the right to disclose personal data to third parties would not give rise to a reasonable expectation of privacy in the communications shared by the Defendant. On the question of the application of the attorney work-product doctrine, the Court suggested that materials prepared by a client using an AI service at the direction of an attorney might suffice to invoke the attorney work-product doctrine, Claude itself disclaims any ability to provide legal advice.

What this Means for Clients Using AI

In both civil and criminal cases in the United States, all parties to the lawsuit are required to keep any evidence related to their claims and subject to some restrictions, are similarly required to share that evidence with the opposing party in the case. Two of the most useful restrictions on this principle are attorney-client privilege and attorney work-product, which generally prevent the opposing party from finding out what clients say to their attorneys and what information their attorneys prepare in the course of representing their clients. The work-product privilege can also extend to materials prepared by clients at the direction of their attorneys in anticipation of litigation. These privileges, however, are not absolute; if the client or attorney is careless in how they share information relating to their case, a Court may find that the privilege no longer applies to the materials.

Clients in all cases should exercise extreme caution in using third-party AI platforms to evaluate their legal claims or summarize case materials. The nature of many, if not all of these platforms involves relinquishing privacy rights to information submitted, and the use of such platforms can create opportunities for opposing parties to discover legal theories, strategies, and mental impressions of the clients and their attorneys.

In our practice, we’ve seen individuals seeking legal advice attempt to use these third-party services to “clean-up” or explain the legal claims they are seeking assistance with. Rest assured, every attorney is trained in what is called issue-spotting. Starting from the first year of law school, up through the Bar Exam, and in everyday practice, we sort through complex fact patterns to identify all potential legal claims and issues to understand your case. In this context, using AI to try to summarize your facts or identify your claims makes our job harder, not easier. In addition to the privacy concerns, AI platforms can suffer from tunnel vision, excluding facts that may be relevant to additional or alternative legal claims or theories of recovery available in the interest of giving you the clean narrative you seek. If you have concerns about your rights being violated in the workplace, call an attorney – not ChatGPT.