Prompts Generated by an Expert Are Discoverable, Court Rules: EDiscovery Case Law

Key Takeaways

• Court finds AI prompts part of expert methodology, thus discoverable
• Rule 29 agreements must be “quite clear” to limit such discovery
• Non‑compliance may trigger Rule 37(b) sanctions for false statements
• Parties must revise Rule 33/34 responses by June 1, 2026

Pulse Analysis

The Connecticut ruling marks a pivotal moment for eDiscovery practitioners grappling with artificial‑intelligence tools. By treating AI prompts as a component of an expert’s methodology, the court affirmed that Rule 26(b)‑type discovery extends to the algorithms and queries that shape evidence selection. This aligns AI‑driven analysis with traditional expert‑witness standards, ensuring that opposing counsel can scrutinize the logical steps behind data culling, not just the final report.

Equally significant is the court’s stance on Rule 29 discovery‑limiting agreements. Judge Farrish emphasized that such contracts must be "quite clear" to exclude specific categories like AI prompts. Ambiguous language will not shield parties from disclosure obligations, and courts remain willing to enforce broader discovery mandates. The threat of Rule 37(b) sanctions for false claims adds a compliance incentive, prompting litigants to conduct diligent searches and provide truthful responses.

For law firms and corporate legal departments, the decision underscores the need to revise discovery protocols and document‑management policies. Teams should catalog AI prompts, query logs, and related metadata alongside traditional expert notes. Early cooperation with opposing counsel can mitigate disputes and avoid costly sanctions. As AI becomes entrenched in litigation strategy, this case sets a precedent that transparency, rather than secrecy, will govern the admissibility and discoverability of machine‑generated insights.