Recent GenAI Rulings Highlight Challenges with Safeguarding Privilege

In this piece authors Ellen Blanchard, Marc B. Collier, Annmarie Giblin, Susana Medeiros, Ethan Glenn & Susan Linda Ross look into how the use of generative AI tools can complicate questions of attorney-client privilege and work-product protection in litigation. They explain that factors such as whether the AI platform is confidential, how prompts and outputs are generated, and the degree of lawyer involvement can influence whether AI-generated materials are considered privileged during discovery disputes.

The article focuses on the decision in United States v. Heppner, where the United States District Court for the Southern District of New York ruled that documents created by a defendant using the AI tool Claude were not protected by attorney-client privilege or the work-product doctrine. The court found that the defendant had independently used a publicly available AI tool to generate potential defense strategies before sharing them with counsel, and therefore the communications were not between a client and lawyer, were not confidential, and were not made for the purpose of obtaining legal advice.

Judge Jed S. Rakoff emphasized that publicly accessible AI platforms cannot create a privileged relationship and that users have limited expectations of privacy when sharing prompts and outputs with such systems. Because the defendant voluntarily disclosed information to an external AI provider and used the tool without counsel’s direction, the court concluded the materials neither qualified as privileged communications nor as attorney work product. The authors note that the ruling heavily relied on the lack of confidentiality associated with public AI tools.

The authors contrast this outcome with other cases, including Concord Music Group v. Anthropic, where prompts crafted by lawyers were protected as work product because they reflected counsel’s legal strategy. They conclude that organizations should carefully manage how employees use generative AI, favor enterprise tools over public platforms, establish clear governance policies, and ensure that legal teams closely supervise AI use in legal investigations to minimize risks to confidentiality and privilege.

 CONTINUE READING…