TL;DR:
A September 29, 2026 order in Villanueva v. Las Vegas Metropolitan Police Department holds that AI-generated case summaries consulted by an expert are discoverable under Rule 26, while the expert’s questions and prompts to a Gen AI platform and an AI-generated deposition outline remain protected as attorney work product. The ruling draws a clear line between AI-produced outputs that inform opinions and the inputs used to generate those outputs. For trial teams, the decision signals the need to implement explicit AI-use protocols, segregate AI prompts from permissible evidence, and prepare for cross-examination and impeachment of AI-assisted opinions. The order is documented as ECF 112 in the District of Nevada, signed September 29, 2026, and it builds on a broader, ongoing debate about how AI materials are treated in discovery. As courts increasingly wrestle with AI in litigation, counsel should align retention, privilege, and discovery plans with this evolving framework.
What happened and why it matters
Jose Villanueva sued the Las Vegas Metropolitan Police Department in a federal civil rights action and related state-law claims. His expert testified that he used a Generative AI platform to research and draft the expert report. Defendants sought production of the expert’s complete ChatGPT history log. A magistrate judge had ordered disclosure of the AI history, including generated case summaries. District Judge Anne R. Traum’s September 29, 2026 order partially sustained and partially reversed that ruling. The court held that AI-generated case summaries are “facts or data” that must be disclosed under Rule 26, while the expert’s AI prompts and an AI-generated deposition outline are protected as drafts or attorney work product. The order, docketed as ECF 112, relies on the Ninth Circuit’s broad reading of “facts or data” and contrasts AI outputs with the inputs used to generate those outputs. The full text confirms the specific holdings and the rationale. (cases.justia.com)
This Nevada decision sits at the cutting edge of a nationwide debate about AI in litigation. A contemporaneous analysis from Mayer Brown highlights that, within months, courts have split on whether AI prompts are discoverable and whether AI-generated outputs used to form opinions must be produced. Villanueva marks a significant tilt toward disclosing AI-derived case material that directly fed into opinions, while shielding the prompts and internal drafting tools that reveal the expert’s thought process. This is not a settled rule across all courts, but it is a concrete milestone in the evolving framework for AI in discovery. (mayerbrown.com)
Practical implications for trial teams
AI-generated case summaries are likely discoverable. Counsel should anticipate that AI-produced summaries, compendiums, or other AI-derived materials actually consulted by an expert and used in forming opinions may need to be disclosed under Rule 26. This requires a careful audit of what AI outputs fed into an expert’s conclusions and where those outputs originated. For many cases, this means producing AI-generated case summaries or similar outputs that were used to support expert opinions. The Villanueva ruling emphasizes that such materials can be treated as “facts or data” in the discovery process. (cases.justia.com)
AI prompts and deposition outlines are protected. Prompts, questions posed to the AI, and AI-generated deposition outlines remain shielded as drafts or attorney work product. This preserves a meaningful boundary between the expert’s computer-assisted inputs and the mental impressions or strategy behind the opinions. Practitioners should log and segregate AI prompts and deposition outlines on privilege logs, ensuring that any production focuses on AI-derived outputs that inform opinions rather than the inputs themselves. (cases.justia.com)
Separate AI uses by category. The Nevada court’s approach suggests a useful framework: (1) AI used for research and to generate case summaries that informed opinions; (2) AI prompts and questions used to guide thought processes; (3) AI-generated deposition outlines used for preparation. Distinguishing these categories in discovery strategy and privilege logs can help manage spoliation risk and avoid over-disclosure. This aligns with guidance from the broader AI-discovery discourse. (cases.justia.com)
Prepare for impeachment and cross-examination of AI-assisted opinions. Even when prompts and the deposition outline are protected, the AI-generated outputs that informed the expert’s opinions may be subject to disclosure and later impeachment at trial. Litigants should plan cross-examination strategies around the reliability and provenance of these AI-derived inputs, including addressing potential hallucinations or miscitations. The opinion’s reasoning and the underlying AI literature become focal points in litigation strategy. (cases.justia.com)
Develop explicit AI-use protocols for experts. In light of Villanueva, firms should require experts to document AI usage in retention letters and discovery stipulations, clearly separating what the expert consulted (case law summaries, datasets, statutes) from what the expert generated (drafts, notes, questions). Rule 26(f) conferences should address AI usage, data sources, and expectations for disclosure and protection. This proactive approach will help prevent widespread disputes later in discovery. (cases.justia.com)
Consider privilege-log discipline and protective orders. Given the protective treatment of AI prompts and deposition outlines, counsel should structure privilege logs to separate AI inputs from AI-derived outputs that may be discoverable. Protective orders may be tailored to preserve the confidentiality of the mental impressions or strategy embedded in prompts while permitting disclosure of the AI-derived materials that fed the expert’s opinion. (cases.justia.com)
How to translate this ruling into immediate steps for practice
Audit prior and current expert materials for AI usage. Identify all AI-derived outputs that informed opinions and prepare to disclose them under Rule 26. Create a clear map from input AI materials to the resulting opinions to facilitate any required production.
Build rigorous AI-use protocols for experts. Require retention letters and engagement agreements that specify which AI tools were used, what data was consulted, and how outputs were incorporated into the report. Prepare a contemporaneous log tracking AI-derived inputs and their role in forming opinions.
Update privilege logs and discovery plans. Log AI prompts and deposition outlines separately from AI-generated case summaries and other outputs. Prepare to argue that prompts and outlines are protected while outputs that informed opinions are discoverable.
Plan for trial-ready responses. Develop cross-examination and impeachment materials focused on the provenance and reliability of AI-assisted content, including potential hallucinations in AI outputs. Use realistic trial simulations to practice objections and openings to AI-driven evidence, leveraging evidence-training resources like Objection Academy to sharpen objection handling and courtroom readiness.
Relevance to ongoing AI practice and training
Villanueva demonstrates that courts are actively shaping the boundaries of AI in litigation, with a practical emphasis on which AI materials become evidence and which remain protected. The decision complements broader commentary from leading firms that federal courts are splitting on AI prompts versus outputs, underscoring a rapidly evolving landscape. For trial teams, this creates a compelling reason to normalize AI governance in discovery and trial readiness. Training resources that focus on objection handling, trial-readiness drills, and expert-cross examination in AI-enabled contexts—such as Objection Academy—remain highly relevant as practitioners translate these rulings into day-to-day litigation workflows. (mayerbrown.com)
Next steps for practitioners
Monitor for any appellate developments in Conservation Law Foundation v. Shell Oil Co. and related AI-prompts disputes, as courts continue to refine the default rules governing AI in discovery. The Nevada decision may be cited by other jurisdictions as they fashion their own precedents. (mayerbrown.com)
Schedule Rule 26 conference discussions with a concrete AI protocol agenda. Include a proposed privilege log schema, an AI-use disclosure plan, and a protective-order framework to manage AI materials consistently across cases.
Consider integrating AI-use governance into trial-readiness programs. Leverage practical obstacles and scenarios highlighted by Villanueva to train attorneys in real-world objections, cross-examinations, and trial simulations that address AI sources, AI-generated outputs, and the limits of AI in expert reasoning.
Explore Objection Academy’s resources for developing objection strategies and cross-examination techniques tailored to AI-derived evidence. Their training aligns with the need for disciplined handling of AI-influenced materials in the courtroom. (cases.justia.com)
Sources:
- Villanueva v. Las Vegas Metropolitan Police Department, et al., Order on ECF Nos. 70, 71, 72 (D. Nev. Sept. 29, 2026) (ECF 112), addressing discovery of an expert’s ChatGPT history log, case summaries, deposition outlines, and drafts. (Justia Dockets & Filings) (cases.justia.com)
- Nevada District Court Splits on AI Prompts in Expert Discovery: Villanueva v. LVMPD (Sept. 29, 2026). Objection Academy article summarizing the ruling and practical implications. (objectionacademy.com)
- Federal Court Splits On Disclosure Of Expert Witness AI Prompts. Mayer Brown, October 9, 2026. Analysis of Villanueva and related cases in the broader circuit landscape. (mayerbrown.com)