• Atlanta
• Boston
• Dallas
• Ft. Lauderdale
• London
• Minneapolis
• New York
• Oakland
• Philadelphia
• Washington, DC

by Jennifer Gibbs

In February 2026, two federal courts issued rulings that together illustrate the unsettled landscape of privilege and work-product protection for AI-generated materials in litigation. In United States v. Heppner, No. 25 Cr. 503 (JSR) (S.D.N.Y. Feb. 17, 2026), Judge Jed S. Rakoff held that documents a criminal defendant generated using the consumer version of Anthropic's Claude AI were protected by neither the attorney-client privilege nor the work-product doctrine. In Warner v. Gilbarco, Inc., No. 2:24-cv-12333 (E.D. Mich. Feb. 10, 2026), Magistrate Judge Anthony P. Patti reached the opposite conclusion, holding that a pro se plaintiff's use of ChatGPT to prepare litigation materials was protected under the work-product doctrine. Although the decisions appear to conflict, the factual differences between them are instructive — and together they underscore that courts will apply traditional privilege principles to AI use in a highly fact-specific manner.

Heppner: Consumer AI Use Defeats Privilege

Bradley Heppner was indicted on securities and wire fraud charges. FBI agents executing a search warrant seized devices containing approximately 31 documents Heppner had generated using the consumer version of Claude. Acting on his own initiative and without counsel's direction, Heppner had used the AI tool to prepare reports outlining potential defense strategies, incorporating information he had learned through discussions with his attorneys.

Judge Rakoff rejected Heppner's privilege claims on three grounds. First, Claude "is not an attorney," and privilege requires "a trusting human relationship" with a licensed professional — a relationship that cannot exist with an AI platform. Second, Heppner had no reasonable expectation of confidentiality because Claude's privacy policy permits Anthropic to collect user inputs, use them for model training, and disclose data to third parties, including "governmental regulatory authorities." Third, Heppner did not communicate with Claude "for the purpose of obtaining legal advice" — he used the tool on his own volition, and Claude itself expressly disclaims providing legal advice.

On work product, the court concluded that the documents were not "prepared by or at the behest of counsel" and drew a critical distinction between materials that merely "affect" counsel's strategy and those that "reflect" counsel's mental impressions at the time of creation. Notably, Judge Rakoff left the door open, suggesting the result might differ if counsel had directed the client to use the AI tool.

Warner: AI as a Tool, Not a Third Party

Warner arose in the context of a pro se employment discrimination lawsuit. The defendants moved to compel production of "all documents and information concerning [plaintiff's] use of third-party AI tools in connection with this lawsuit."

Magistrate Judge Patti denied the motion on multiple grounds. The court held that Warner's AI-assisted litigation materials were protected by the work-product doctrine and, critically, rejected the argument that using ChatGPT constituted disclosure to a third party that would waive that protection. The court stated plainly that "ChatGPT (and other generative AI programs) are tools, not persons, even if they may have administrators somewhere in the background." The court further found that the defendants' theory "would nullify work-product protection in nearly every modern drafting environment, a result no court has endorsed," and characterized the request as a "fishing expedition" untethered from Rule 26 relevance.

Implications for Litigation

Despite their seemingly opposite outcomes, Heppner and Warner can be reconciled. In Heppner, the defendant acted unilaterally and without counsel's involvement, used a consumer platform with broad data-sharing terms, and the government sought access to already-seized documents. In Warner, a pro se litigant effectively acting as her own counsel used AI as a drafting tool, and the opposing party sought affirmative discovery into that usage.

From these two cases, several takeaways emerge. First, the selection of the AI tool matters: enterprise platforms with contractual confidentiality protections present a materially different risk profile than consumer tools whose terms of service permit broad data collection. Second, attorney direction is critical — Heppner strongly suggests that AI use at counsel's direction will receive more favorable privilege treatment than a client's unilateral use. Third, organizations should update their AI governance policies to address these risks, including restricting the input of privileged or confidential information into consumer platforms. Finally, litigants should anticipate that opposing parties may increasingly seek discovery into AI usage, and should be prepared to defend against — or strategically pursue — such requests.

These decisions represent an early but significant chapter in the law's effort to apply established privilege principles to rapidly-evolving technology. The critical factors going forward will be the contractual and technical features of the AI platform, the degree of attorney involvement, and whether the materials genuinely reflect litigation strategy prepared in anticipation of litigation.

_________________________________

The opinions expressed are those of the authors and do not necessarily reflect the views of the firm or its clients. This article is for general information purposes and is not intended to be and should not be taken as legal advice.