By Denise Chicoine and Andrew Oringer
It has become axiomatic to say that artificial intelligence (AI) is everywhere. As the world is changing, the bar organizations and associations that have an interest in regulating the legal profession are trying to chase the moving target that AI presents. Some of the issues, such as those involving inadequate review of and supervision regarding AI output, are arguably fairly intuitive and relatively straightforward. Others, like the risks that AI might hallucinate (in the current vernacular) fictional cases or that confidential information plugged into an AI platform might be revealed by the platform in an effort to answer other users, are arguably less intuitive and more subtle and nuanced.
Further to this latter point, there is the emerging related issue of whether the use of AI, whether by a law firm or a client, could result in the loss of the attorney-client privilege for the information submitted as a part of an AI query. For example, in United States v. Heppner, the federal trial court for the Southern District of New York ruled that when a defendant communicates with a publicly available AI platform in connection with a pending criminal investigation, the communications are not protected by the attorney-client privilege or the attorney work-product doctrine.
Background
The attorney-client privilege shields from disclosure in legal proceedings any oral, written, or electronic communications made in confidence for the purpose of seeking or providing legal advice. Courts construe the attorney-client privilege narrowly because it operates as an exception to the rule that “all relevant proof is essential” for a complete record and for “confidence in the fair administration of justice.” See, e.g., In Re Six Grand Jury Witnesses, 979 F.2d 939 (1992). The work-product doctrine preserves a zone of privacy in which a lawyer can prepare and develop legal theories and strategies with an eye toward litigation, free from discovery by adversaries. See, e.g., United States v. Nobles, 422 U.S. 225, 238 (1975).
The ruling in Heppner addresses a question that may well be of first impression nationwide: how do these doctrines apply in the context of the use of AI in litigation?
Facts of Heppner
In Heppner, the defendant was indicted for securities fraud, wire fraud and related charges. The criminal charges stemmed from actions the defendant allegedly took as an executive of a publicly traded company.
The defendant used a highly developed AI tool after he was subpoenaed by a grand jury and became aware he was the target of a criminal investigation. Defense counsel did not ask the defendant to perform the AI search.
Thereafter, FBI agents arrested the defendant and executed a search warrant at his home. During the course of the search, the government seized numerous documents and electronic devices. Defense counsel claimed that some of the seized documents that memorialized the defendant’s communications with the AI tool were privileged, on the basis that the defendant used AI for the purpose of communicating with his lawyers to outline a strategy in response to likely charges.
The Ruling in Heppner
Heppner rejects defense counsel’s argument that the materials in question were protected from disclosure because the defendant had input information to AI that he had learned from counsel, created the AI output for the purpose of obtaining legal advice, and later shared that output with counsel.
The court said that all recognized privileges require, among other things, “a trusting human relationship,” such as, in the attorney-client context, a relationship “with a licensed professional who owes fiduciary duties and is subject to discipline. No such relationship exists, or could exist, between an AI user and a platform such as” the one used in Heppner.
The court also stated that there was no reasonable expectation of confidentiality in communications with the AI tool. The AI platform has a written privacy policy to which users must consent which provides that all data collected is used to “train” the AI platform and could be disclosed to a host of third parties, including governmental regulatory authorities. (Lawyers are well-advised to take note of this aspect of the way in which AI may be trained.)
The court noted that the defendant did not communicate with the AI tool for the purpose of obtaining legal advice. He communicated with the platform of his own volition, and “what matters for the attorney-client privilege is whether [the defendant] intended to obtain legal advice from [the AI platform], not whether he later shared [the AI platform’s] outputs with counsel.”
In regard to the work-product doctrine, the court stated that the purpose of the doctrine “is not generally promoted by shielding from discovery materials in an attorney’s possession that were prepared neither by the attorney nor his agents.” The court, which is in the Second Circuit, pointed out that the Second Circuit has repeatedly stressed that the purpose of the doctrine is to protect lawyers’ mental processes. The court held that the AI documents do not merit protection under the work-product doctrine because, even assuming they were prepared “in anticipation of litigation,” they were nevertheless not prepared by or at the behest of counsel and did not reflect defense counsel’s strategy.
Heppner concludes: “AI’s novelty does not mean that its use is not subject to longstanding principles, such as those governing the attorney-client privilege and the work product doctrine.” Heppner may be appealed, but, regardless of where the case ends, it presents a cautionary tale regarding confidentiality and privilege.
Conclusion
The world is continuing to evolve regarding the development of AI. Among the legal ethics issues that will continue to emerge are those relating to confidentiality and privilege, as Heppner starkly illustrates. Attorneys, and indeed their clients, should pay heed to when the use of AI – not only with respect to the absence of privilege in their interactions with AI platforms (as Heppner directly shows) but also with respect to whether the inputting of otherwise confidential information will be used by the platform in a way that discloses the information to others (as noted in Heppner) – may have unforeseen legal consequences.
If you have any questions about Heppner, the attorney-client privilege, the attorney work-product doctrine or any other ethical issues affecting attorneys, please do not hesitate to contact our dedicated Ethics Group or Employment Group.
