In what appears to be an issue of first impression, on February 10, 2026, Judge Rakoff of the U.S. District Court for the Southern District of New York ruled from the bench in the case of United States v. Heppner that a criminal defendant’s written interactions with a publicly available generative AI platform are not protected by the attorney-client privilege or the work product doctrine.
Background
A grand jury indicted the defendant on October 28, 2025, for securities fraud, wire fraud, conspiracy, making false statements to auditors, and falsifying corporate records. In connection with defendant’s arrest, the FBI executed a search warrant at his home and seized documents and devices, including approximately thirty-one “AI Documents” memorializing his 2025 communications with Claude after he had received a grand jury subpoena and understood he was a target. Defense counsel asserted privilege on the grounds that the defendant input information learned from counsel, prepared the materials to speak with counsel to obtain legal advice, and later shared the contents with counsel. The defendant’s attorneys conceded that they did not direct the defendant to use Claude. Pursuant to a privilege protocol.
Holding
Judge Rakoff held that the AI Documents are not protected by attorney-client privilege because they are not communications with an attorney, were not confidential, and were not made for the purpose of obtaining legal advice from an attorney. Judge Rakoff further held that the AI generated documents are not protected under the work product privilege because they were not prepared by or at the behest of counsel and did not reflect counsel’s strategy at the time they were created.
Judge Rakoff’s opinion marks an early but consequential application of traditional privilege and work product principles to generative AI. By grounding the analysis in the absence of a human professional relationship, the lack of confidentiality arising from platform terms, and the requirement of attorney direction for work product, the Court signals that AI’s novelty does not expand evidentiary protections.
Conclusion
Generative AI platforms are being utilized more and more by consumers to, among other things, provide advice or summarize information. Attorneys need to be proactive in helping their clients understand the ramifications of using AI tools in disputes. Attorneys should warn clients against unilateral AI consultations regarding pending matters, and establish clear protocols for any technology-assisted work product that ensure attorney direction, confidentiality, and control.
SOURCE: U.S. v. Heppner, Case No. 1:25-cr-00503-JSR, Dkt 27.
Patrick J. “P.J.” Gallo is a seasoned civil and commercial litigation attorney with extensive experience representing businesses, entrepreneurs, and individuals in complex disputes and high-stakes litigation. He provides strategic, results-driven advocacy both in and out of the courtroom. Clients rely on P.J. for practical legal strategies that protect their operations, reputation, and financial interests. To contact Patrick, visit macelree.com/contact-us.
