A trailblazing decision has been issued by Judge Rakoff of the United States Federal Court for the Southern District of New York. The decision, issued on February 17, 2026, analyzes the applicability of attorney-client privilege and the work product doctrine to third-party generative Artificial Intelligence (AI) platforms.
In U.S. v. Heppner, 25 Civ. 503, the Court answers a question of first impression nationwide—are a user’s communications with a publicly available AI protected by attorney-client privilege or the work product doctrine? The answer, at least on the facts of that case, was no.
Background
After receiving a grand jury subpoena in connection with an investigation into, among other things, suspected securities fraud, Defendant Bradley Heppner utilized Claude[1] and, through different queries, prepared documents outlining his defense strategies and developing legal and factual arguments in anticipation of a potential indictment. Importantly, Heppner’s inquiries were not conducted under the direction of any counsel but were of his own initiative.
In connection with his indictment and arrest, the FBI executed a search warrant on Heppner’s home that resulted in the seizure of documents and materials memorializing Heppner’s communications with Claude. Heppner’s attorneys subsequently asserted that these documents were privileged.
The Court’s Holdings
The Court held that the attorney-client privilege did not apply because Claude was indisputably not an attorney. The Court also reasoned that there was no expectation of confidentiality because Anthropic’s privacy policy allowed it to collect data on user “inputs” and “outputs” and reserved the right to disclose that information to third parties. The Court went further and held that Heppner’s communications were not for the purpose of seeking legal advice because Claude disclaims providing legal advice and, in response to queries regarding legal advice, states “I’m not a lawyer and can’t provide formal legal advice or recommendations.”
In a similar vein, the Court held that the documents were not protected by the work product doctrine because, even if they were prepared in anticipation of litigation, they were not prepared by or at the behest of counsel.
Why This Matters
The Court’s decision, while specific to the facts of this case, is significant for several reasons:
- It is one of the first decisions to address issues of privacy, privilege, and confidentiality as applied to AI in the litigation context and establishes analytical precedent for similar situations moving forward;
- It demonstrates that, despite the novelty of AI, Courts and litigants are actively mapping out the advantages and pitfalls of its usage in legal applications;
- It raises new and important questions about how attorneys approach litigation, and even everyday interactions with clients, in this modern age; and
- It reinforces that, at this early stage, AI when used for legal issues is best used in conjunction with and as a supplement to legal advice from an attorney, and, even then, only at the attorney’s direction.
If you have any questions about this decision or its potential impact on your business or practice, please contact the authors, the Garfunkel Wild attorney with whom you regularly work, or contact us at [email protected].
* * * * *
[1] A generative AI platform operated by Anthropic.
