TL; DR: A recent US court ruling has established that communication between an individual and AI tools are not protected by privilege or work product doctrine, even if they are related to legal matters or are created as part of planning a claim or defence. The content of these communications can be considered discoverable and admissible as evidence, which can seriously jeopardize your legal position. Discuss legal matters only with a qualified legal representative to ensure confidentiality.
Artificial intelligence can be an incredibly helpful tool in many areas. However, with respect to the law, consumer AI tools are not a replacement for qualified legal advice. Many AI models are known for creating cases that don’t exist, or incorrectly interpreting cases that do. However, recent developments have made it especially important to be aware that your conversations with AI may not be protected by any privacy or confidentiality doctrine, even if they are directly related to legal proceedings.
A February 17, 2026 decision from the Southern District of New York, United States v. Heppner (2026 WL 436479) provides a stark warning against the dangers of using AI tools for a legal purpose. Mr. Heppner was charged with securities and wire fraud, as well as conspiracy and other related offences. After learning he was the subject of a federal investigation Mr. Heppner engaged a consumer AI model for advice on approximately thirty-one inquiries including potential defence strategies, and what arguments could be made based on the facts that he provided. Mr. Heppner’s electronic devices were seized as part of the investigations, and the prosecution became aware of the communication between Mr. Heppner and the AI model.
Mr. Heppner’s counsel argued that these communications were inadmissible as evidence against Mr. Heppner. They contended that the information provided to, and received from, from the AI model was protected by lawyer-client privilege, and that the inquiries constituted work product as they related to his pending charges and associated defence. In support of their position, they argued that he had a) input information provided to him by his counsel, b) created the AI communications for the purpose of speaking with counsel to obtain legal advice, and c) shared the AI communications with counsel for the purpose of developing a defence.
The court ruled that these communications were not subject to lawyer-client privilege.
In order to invoke that protection the material must be confidential communication between a lawyer and a client for the purpose of obtaining legal advice. The conversations in question were between Mr. Heppner and the AI model, and not a lawyer, which means that the elements of privilege were not met.
The court further ruled that these communications were not protected as “work product”.
The work product doctrine provides qualified protection for materials prepared by, or at the behest of, counsel in anticipation of litigation or for trial. As these communications were not created by counsel, nor at counsel’s request, the court refused to apply the doctrine to the communications.
As such, all the information that Mr. Heppner provided to the AI model as part of preparing a defence – including sensitive or confidential information that may significantly damage his legal position – were determined to be admissible as evidence against him in his criminal proceedings.
Be very cautious of any use of AI with respect to legal matters. Assume that any conversations you have with consumer AI platforms such as ChatGPT, Claude, Grok, or other AI tools are not confidential and may be considered discoverable by an opposing party, and admissible as evidence against you. AI is many things, but “lawyer” is not one of them – in order to protect yourself and your interests, ensure that you are discussing legal matters and obtaining legal advice only with your legal representative.
Written by Jim Rudy and Josh Valler