HaystackID: Protecting Privilege and Work Product in Discovery After Heppner and Warner

Key Takeaways

• Heppner denies privilege for AI content not lawyer‑directed
• Warner extends work product protection to AI‑generated plaintiff materials
• Counsel must supervise AI tools to maintain privilege
• Service‑provider agreements essential for controlled AI document review

Summary

Recent rulings in United States v. Heppner and Warner v. Gilbarco illustrate how courts are grappling with the intersection of generative AI and evidentiary protections. Heppner held that AI‑generated content, created without direct attorney instruction, is not shielded by lawyer‑client privilege or work‑product doctrine in a criminal case. Conversely, Warner found that AI‑created materials for a pro se plaintiff in an employment discrimination suit are covered by the work‑product doctrine. Together, the decisions underscore the need for lawyers to tightly control AI use during discovery.

Pulse Analysis

Generative AI is rapidly becoming a staple in e‑discovery, offering speed and cost efficiencies for reviewing massive data sets. Yet, as firms integrate these tools, they must recognize that traditional evidentiary doctrines were crafted before such technology existed. Courts are now testing whether privilege and work‑product shields automatically apply to AI‑produced outputs, forcing practitioners to reassess risk management strategies and client communication.

In United States v. Heppner, the district court concluded that AI‑generated responses to a defendant’s prompts—absent any attorney direction—did not merit lawyer‑client privilege or work‑product protection. The ruling emphasized that privilege hinges on the attorney’s control over the communication, not merely the content’s relevance. This decision alerts criminal defense teams that unsupervised AI usage can inadvertently waive confidentiality, potentially exposing critical evidence during discovery.

Warner v. Gilbarco took a contrasting stance, holding that AI‑created documents prepared for a pro se plaintiff were protected as work product in an employment discrimination case. The court recognized the plaintiff’s reliance on AI as a substitute for legal expertise, extending the doctrine to preserve strategic materials. The split between Heppner and Warner highlights the importance of establishing clear protocols: firms should draft service‑provider agreements, enforce attorney‑led prompts, and document supervision of AI tools to safeguard privilege across both civil and criminal matters.