A Warning for Clients: Using AI for Legal Strategy May Not Be Privileged

Most of us have heard of Artificial Intelligence (“AI”) tools such as ChatGPT and Claude. AI can be a powerful tool if used correctly and under the right conditions. Unfortunately, it can also be misleading, and as a new Federal court decision informs us, its use can result in significant negative consequences, including the loss of the attorney-client privilege. In our practices, we are seeing exponential growth in clients using AI in numerous ways without understanding the significant risks that can arise from using these products for legal purposes. This article serves as a cautionary tale for our clients and attorneys.

In United States v. Heppner, No. 1:25-cr-00503-JSR (S.D.N.Y. Feb. 17, 2026), the U.S. District Court for the Southern District of New York ruled in a case of first impression that communications between a criminal defendant and the Claude AI platform are not protected by attorney-client privilege or the work product doctrine. The court held that AI communications lack the essential elements of both legal protections: there is no attorney-client relationship between a user and an AI platform, the communications are not confidential due to the AI company's data collection and sharing policies, the defendant did not seek legal advice from the AI (which explicitly disclaims providing legal counsel), and the documents were not prepared by or at the direction of counsel.

Unbeknownst to his attorney, defendant Bradley Heppner used Claude, a generative AI platform, to analyze facts and prepare strategy memos, in part based on his communications with his attorney. The Court ruled that the client’s inquiries and the resulting memoranda were not protected by the attorney-client privilege or the attorney work product doctrine, even though Heppner and his attorney argued the inquiries and memoranda were meant to be used in their own discussions, which would normally be protected by the attorney-client privilege.

The Court held that Heppner’s communications with Claude were communications between two non-attorney entities, and therefore, they were not protected, despite the topics being generated during attorney-client communications. The fact that the attorney was unaware of these actions and did not ask his client to engage in these inquiries was not an acceptable defense.

There are several key takeaways from this decision for both clients and attorneys:

• Clients should consult with their attorneys before embarking on their own AI research so that their attorneys are aware of this potential and can advise the client accordingly.
  
  
• If clients have engaged in such research before retaining or consulting an attorney, they should advise their attorney of the research and the specific inquiries.
  
  
• Attorneys should proactively ask clients whether they have used any generative AI tools so that they may assess the potential risks.
  
  
• Attorneys and legal professionals should refrain from entering confidential details into open-source generative AI tools for their own research.

If you have any questions about the use of AI, whether for legal or business reasons, our attorneys are here to help you navigate the ever-changing landscape to ensure that you and your businesses are protected.

Note: DeWitt LLP attorneys and staff do not use open-source Generative AI tools such as Claude or ChatGPT.  Instead, we use a closed Generative AI platform that combines the power of AI with the confidentiality of an internal system. This ensures the confidentiality of our research and communications. 

Author

Michele L. Perreault
Partner
Madison, WI