Client Use of AI is Not Protected by Privilege
NY Court Holds The Work Product Doctrine and Attorney-Client Privilege Do Not Apply
Exchanges with a generative artificial intelligence platform are not protected by the work product doctrine or attorney-client privilege according to a recent decision by the U.S. District Court for the Southern District of New York. This recent decision acts as a warning to anyone using AI platforms in other jurisdictions that their interactions are likely not protected by privilege or the work-product doctrine. United States v. Heppner
The specific question addressed by the court was whether a user’s communications with a publicly available AI platform related to a criminal investigation are protected by attorney-client privilege or the work product doctrine. The court has clearly stated that “the answer is no.”
Defendant’s counsel argued that the communications with “Claude” were prepared by the Defendant in anticipation of a criminal indictment and attempted to assert privilege. The Attorney-Client Privilege protects communications between a client and their attorney that are intended to be and are kept confidential for the purpose of obtaining legal advice. Such protection prevents the communication from being disclosed to the opposing side in a criminal or civil matter. The Court determined that the generative artificial intelligence platform is not an attorney such that communications with it are not protected by privilege. The Court went on to state that the use of other internet-based software is also “not intrinsically privileged”. furthermore, Claude’s privacy policy maintains a right to disclose data to third parties such as governmental regulatory authorities. Users do not have a substantial privacy interest in conversations with AI platforms that retain such data. Lastly, the Defendant did not communicate with Claude for the purpose of obtaining legal advice because he could not have obtained legal advice from an AI platform which is not a licensed attorney.
Although this decision specifically affects those in New York and focuses on criminal matters, the same logic might apply to courts in Pennsylvania and other states that are asked to rule upon the protections of communications with AI. According to the court, this is a matter of “first impression” which means that they are not aware of any other court addressing the same question.
The safest approach for clients in the absence of decisions to the contrary from other courts is likely to avoid the use of generative AI platforms in relation to any active, pending, or possible legal dispute whether civil or criminal. Any such use is likely to be discoverable by the opposing side.
For help with litigation matters in South Central Pennsylvania, contact Attorney Hunter Schenck at hschenck@cgalaw.com or using the CGA online inquiry feature to set up a consultation with Attorney Schenck.
