Source: Phillips Lytle LLP

Open source AI prompts and responses raise unsettled privilege issues in U.S. litigation; some courts have treated such prompts as protected work product (for example, Morgan v. V2X, Inc., D. Colo., March 30, 2026), while at least one federal court in New York held in a criminal case that prompts and responses are not protected attorney-client communications or privileged work product (United States v. Heppner, S.D.N.Y., Feb. 17, 2026). Those inconsistent rulings create a significant risk that prompts fed to open source AI could be discoverable, potentially revealing litigation strategies or weaknesses. Because opposing parties may seek AI prompts and responses in discovery, users must consider preservation, collection, review, and the possibility of production under the governing law.

AI is playing an increasing role in litigation work, and many litigators are exploring closed-source or proprietary AI platforms designed for lawyers that the article says may better preserve privilege and confidentiality and help reduce litigation spend. By contrast, open source AI offers none of these assurances; counsel cannot rely on open source AI analyses without rechecking them for completeness, accuracy, and hallucinations, a process that the article warns will be borne by the client and may raise costs. The piece advises contacting litigation counsel first to ask which AI platforms they use, how client data is protected, and how AI will be employed to increase efficiency before consulting non-confidential open source AI tools. The author is Jacob S. Sonner, Partner at Phillips Lytle LLP and a member of the firm’s Litigation Practice Group.