BEWARE OF DANGERS IN SHARING CONFIDENTIAL ATTORNEY INFORMATION ON GENERATIVE AI

A Federal Court in the Southern District of New York recently ruled in United States v. Heppner that a client who used his attorney’s suggestions, legal analysis and/or printed materials in making an inquiry using a generative artificial intelligence (“GAI”) program (which includes applications such as ChatGPT, Google Gemini and Claude) waived the attorney-client privilege and the work product privilege of confidentiality as to the responses received from the GAI application.

As a result of the Court’s ruling, opposing counsel was entitled to the documents that the client received from the GAI program. The case is a caution that AI-generated materials prepared by a client (as opposed to such work prepared by an attorney) are generally not shielded by legal privilege.  It serves as a practice pointer to attorneys to caution clients against “sharing” attorney input with generative AI, and a warning to clients not to do this!

For a detailed analysis of the case from the Harvard Law Review, click here.

By Douglas J. Sanderson