TL;DR: A Nevada federal district court order in Villanueva v. Las Vegas Metropolitan Police Department holds that AI-generated case summaries relied on by an expert in forming opinions are discoverable under Rule 26, while the expert’s prompts to the AI, deposition outline, and AI-generated draft outputs remain protected as trial preparation or expert work product. The Sept. 29, 2026 ruling clarifies the divide between what must be produced and what can stay confidential, signaling a developing split across jurisdictions on AI-assisted expert discovery. Expect many litigators to push for AI-output disclosures, while defense teams seek to shield prompts and drafting material. The court’s decision also emphasizes careful protocol for using generative AI in expert work and the ongoing risk of AI hallucinations in legal research. This is a timely development in an active area of AI-evidentiary practice in federal courts across the United States. (Villanueva v. Las Vegas Metropolitan Police Department, No. 2:24-cv-00125-ART-DJA, D. Nev., Sept. 29, 2026; ECF No. 112) (docs.justia.com)
What happened
In Villanueva v. Las Vegas Metropolitan Police Department, the plaintiff’s expert relied on ChatGPT to generate case-law summaries that informed his expert opinion on an excessive-force claim. After a contested discovery dispute, District Judge Anne R. Traum issued an order on September 29, 2026 (ECF No. 112), clarifying what AI-related materials must be disclosed under Federal Rules of Civil Procedure Rule 26 and what remains protected. The court found that the AI-generated case summaries themselves constitute “facts or data” considered by the expert and therefore must be disclosed under Rule 26(a)(2)(B)(ii). By contrast, the court held that the expert’s questions and prompts to ChatGPT, the AI-generated deposition outline, and the portions of the log that reflect the expert’s drafting process are protected either as trial preparation materials or as draft-work product, and need not be produced. The prior bench ruling by Magistrate Judge William G. Albregts, on December 23, 2025, granting production of the ChatGPT history log, serves as a backdrop to the current decision. The Nevada court thus draws a line between material that informs the expert’s opinion and material that reveals the expert’s mental impressions or trial strategy. (docs.justia.com)
The court’s analysis relies on the Federal Rules of Civil Procedure and established Ninth Circuit authority about what counts as “facts or data considered” by an expert (Rule 26(a)(2)(B)(ii)) versus protected draft material and trial preparation (Rules 26(b)(3) and 26(b)(4)(B)). It cites Republic of Ecuador v. Mackay for the broad interpretation of “facts or data considered,” while also distinguishing items that reveal the expert’s mental impressions. Importantly, the order confirms that AI-generated content that feeds into the expert’s opinion—such as case-law summaries—falls within the scope of discoverable material, whereas other AI outputs that merely structure or strategize the expert’s work remain privileged. The court expressly notes that this remains a developing area with jurisdictional splits, and oral argument on related objections was scheduled for November 3, 2026. (docs.justia.com)
Why it matters for trial lawyers
Discovery leverage shifts with AI outputs. If opposing experts have used AI to research and generate case summaries, those summaries may be discoverable and could become a key evidentiary anchor in cross-examination and impeachment. Litigants should anticipate requests for the AI-generated case summaries that informed an expert’s opinion and be prepared to produce those outputs or seek protective orders as appropriate. The Villanueva decision makes clear that not all AI-derived content is automatically public, but the summaries themselves are subject to disclosure. (docs.justia.com)
AI prompts and drafting remains privileged. The court’s ruling draws a bright line around prompts, deposition-outline prompts, and AI-assisted drafting outputs, keeping those materials protected as potential mental impressions or trial preparation. This matters for strategy because it preserves some of the expert’s internal processes from discovery, while still enabling parties to probe the substantive inputs that shaped the expert’s conclusions. Practitioners should design discovery requests that target the discoverable outputs (the case summaries) while recognizing the protected nature of prompts and drafting records. (docs.justia.com)
Signals a jurisdictional split to watch. The Nevada ruling follows a line of decisions showing mixed treatment of AI prompts and related materials across circuits and districts. This underscores the importance of jurisdiction-specific discovery planning and preservation protocols, and it suggests that parties should be ready to tailor arguments to their forum’s approach to AI in expert discovery. Counsel should monitor related developments, including ongoing objections or appeals in Conservation Law Foundation v. Shell Oil Co. and similar debates elsewhere. (docs.justia.com)
Practical implications for practice management. Firms should implement clear protocols for AI use by experts, including documenting which AI outputs feed into opinions, how those outputs are reviewed, and where case-summaries sit in the evidentiary chain. The Nevada order reinforces the value of keeping the AI-generated case-summaries in a well-documented, reviewable form that could be produced upon request, while protecting internal prompts and drafting aids. Such protocols help avoid inadvertent waiver of privileges and streamline future discovery disputes. (docs.justia.com)
Risk management for AI-generated content. The decision also highlights the risk of hallucinated or questionable AI outputs appearing in expert reports. Practitioners should prepare to authenticate AI-derived case summaries with traditional sources and to address potential inaccuracies during cross-examination, rather than accepting AI outputs at face value. This aligns with broader trial readiness practices that emphasize verification, reproducibility, and robust impeachment tools—areas where Objection Academy’s training modules can help teams rehearse objections and filter questions around AI-derived content. (docs.justia.com)
Practical steps for litigators
In Rule 26 conferences, push for production of the AI-generated case summaries that fed into the expert’s opinion, while requesting protective orders for prompts and drafting materials. Prepare to articulate the specific category of material sought under 26(a)(2)(B)(ii) and distinguish it from work product. (docs.justia.com)
Build an AI-use protocol for experts. Include a detailed log of inputs and outputs, specify which AI-derived outputs are treated as facts or data, and define which items are considered draft material or trial preparation. This can help avoid waiver and streamline future disputes. (docs.justia.com)
Prepare for cross-examination on AI-generated content. Expect scrutiny of the accuracy and relevance of AI-produced case summaries, and have a plan to verify sources, address hallucinations, and confront the expert with any gaps between AI outputs and primary authorities. The Nevada decision explicitly warns about potential “hallucinated case citations” in AI outputs. (docs.justia.com)
Monitor developments across jurisdictions. The Villanueva decision is part of a broader debate on AI in expert discovery. Attorneys should track similar orders and any appellate decisions to anticipate forum-specific strategies for AI materials in expert reports, depositions, and trial. (docs.justia.com)
Leverage training resources for trial teams. In parallel with defending or attacking AI-based expert material, trial teams can benefit from targeted objection and deposition training to handle AI-generated evidence, a need that grows with cases like Villanueva. Objection Academy and comparable training can assist in developing effective cross-examination techniques around AI-sourced content while reinforcing proper objection to privileged material. (docs.justia.com)
Bottom line for the current year
Villanueva v. LVMPD marks a concrete, recent development instructing federal courts on how AI-assisted expert work will be treated in discovery. The ruling confirms that AI-generated case summaries used by an expert to form opinions are discoverable as facts or data considered, while AI prompts and other drafting materials remain protected. For trial teams, this translates into a practical playbook: seek the AI-generated outputs that fed opinions, protect the rest, and implement robust AI-use protocols for future cases. As the AI-evidence landscape continues to evolve, disciplined preparation and clear procedural guidelines will distinguish the most effective trial teams in the years ahead.
Sources:
- Villanueva v. Las Vegas Metropolitan Police Department, No. 2:24-cv-00125-ART-DJA, D. Nev., Sept. 29, 2026 (ECF No. 112), Justia docket and order text. (docs.justia.com)
- Mayer Brown, Federal Court Splits On Disclosure Of Expert Witness AI Prompts, Oct. 9, 2026 (analysis of the Villanueva decision and its implications). (mayerbrown.com)