Federal Court Rules Client’s Use of Generative AI Is Not Privileged

Key Takeaways

• The U.S. District Court for the Southern District of New York recently held that communications by a criminal defendant with a publicly available generative AI platform are not protected by attorney-client privilege or work product privilege.
• Applying privilege principles, the court found no attorney-client relationship between the individual and the AI platform, no reasonable expectation of confidentiality, and no protected legal advice.
• Client-generated AI materials created without direction from counsel do not qualify as work product.
• The decision reinforces the importance of taking care regarding use of publicly available generative AI tools by clients for purposes relating to legal matters.

On February 17, 2026, the Southern District of New York, in United States v. Bradley Heppner, held that a criminal defendant's written exchanges with a “publicly available AI platform” are not protected by attorney-client privilege or work product doctrine and, thus, could be inspected by the government. The ruling appears to be the first federal decision squarely addressing privilege claims for communications with a generative AI platform of this type.

The decision confirms long-established privilege principles apply with full force to AI tools and using an AI platform that does not provide confidentiality protections—particularly one that trains on user data—can destroy privilege and other confidentiality claims.

Background

• After receiving a grand jury subpoena and becoming aware that he was the target of a criminal investigation, the defendant—acting on his own initiative without instruction from counsel—used a free generative AI platform to prepare documents outlining his potential defense strategy and responses to potential charges.
• The AI platform he used stated in its privacy policy that it (1) collects user inputs, as well as the platform’s outputs; (2) uses inputs and outputs to further train the platform; and (3) reserves the right to disclose both inputs and outputs to a “host of ‘third parties,’ including ‘governmental regulatory authorities.’”
• The defendant later provided the AI-generated documents to his counsel, who reviewed them.
• In October 2025, a grand jury indicted the defendant on securities fraud and related charges arising from alleged misconduct as a public company executive.
• Upon arrest in November 2025, the FBI executed a search warrant at the defendant’s home and seized materials including the documents generated by the defendant’s use of the AI platform.
• Through counsel, the defendant asserted privilege over the AI-generated documents. The government moved for a ruling that the AI-generated documents were not privileged. The court granted the government’s motion.

Attorney‑Client Privilege: No Protection for Exchanges With 'Publicly Available' Generative AI

The court applied the same three‑part privilege test that would apply to any communication by a client. It considered whether the AI-generated documents were communications (1) between client and attorney, (2) intended and kept confidential, and (3) for the purpose of obtaining legal advice. It found that the AI-generated documents failed all three elements.

No Attorney‑Client Relationship With the AI Platform

• The AI platform is not a lawyer, and no attorney‑client relationship existed between the defendant and the platform. Communications with the AI platform cannot qualify as communications “between a client and his or her attorney.”
• All recognized privileges presuppose “a trusting human relationship” (emphasis added) with a licensed professional bound by fiduciary duties—conditions the AI platform cannot meet.

Lack of Confidentiality Given the Platform’s Data Practices

• The AI platform’s privacy policy disclosed that it collects user inputs and outputs, uses data to train the model, and may disclose data to third parties, including government authorities.
• Users who voluntarily submit content to such systems—knowing the provider retains and may share it—lack any reasonable expectation of confidentiality.

Not Communications for the Purpose of Obtaining Legal Advice From Counsel

• Counsel conceded they “did not direct [the defendant] to run [AI] searches.” Because the defendant acted on his own volition, the court asked whether he sought legal advice from the AI platform—and found the platform’s disclaimer that it “can’t provide formal legal advice” dispositive.
• The defendant’s later sharing of the nonprivileged AI communications with counsel did not retroactively convert them into privileged material. Materials that “would not be privileged if they remained in [the client’s] hands” do not “acquire protection merely because they were transferred” to an attorney.

The court also held that any privilege over information conveyed by counsel to defendant and then included in the defendant’s inputs to the AI platform was waived when the defendant disclosed it to the AI platform—no differently than sharing with any other third party with no expectation of privacy or privilege.

Work Product Doctrine: No Protection for Client‑Generated AI Materials

Even assuming the AI-generated documents were created “in anticipation of litigation,” the court found two independent reasons they did not qualify as work product.

Not Prepared by or at the Direction of Counsel

• Counsel confirmed the AI-generated documents “were prepared by the defendant on his own volition.” Because counsel did not direct the defendant to use the AI platform, he was not acting as counsel’s agent.
• The doctrine is “not generally promoted by shielding from discovery materials in an attorney’s possession that were prepared neither by the attorney nor his agents.”

Do Not Reflect Counsel's Mental Impressions or Strategy

• Counsel acknowledged the AI-generated documents did not “reflect” counsel's strategy at the time of creation. Protection requires a “real, rather than speculative, concern” that disclosure would expose counsel's thought processes.

Implications for Use of Clients’ Use of Publicly Available AI Tools in Legal Contexts

The court stressed that AI’s novelty does not displace settled evidentiary rules:

• Communications by a nonlawyer with an AI platform that trains on inputs and outputs and has the stated right to disclose inputs and outputs to third parties such as the government are neither attorney‑client communications nor confidential communications absent a genuine attorney‑agent relationship and an indication that the client communicated with the AI platform as an agent of counsel.
• Use of AI tools whose providers reserve rights to retain, train on, and disclose user data undercuts any reasonable expectation of confidentiality, which may well waive any attorney-client privilege that might otherwise have attached.
• Client‑generated AI materials, without direction from counsel, will not be treated as work product.
• Nonprivileged AI-generated outputs do not become privileged simply because they are later sent to counsel.

The ruling underscores that long-established rules of privilege apply to generative AI just as they apply to telephone calls, memorandums, and emails. Clients seeking to use generative AI tools in litigation contexts will be well-served to pay close attention to platform privacy policies and other disclosures and whether AI‑assisted work is undertaken by counsel or at counsel’s direction.