Are an Expert’s A.I. Prompts Discoverable? Maybe.

Failure to Follow Court Directive Nearly Snatched Defeat From Victory

October 10, 2026

Published by Michael Berman at October 12, 2026

Tags

Villanueva v. Las Vegas Metropolitan Police Dept., 2026 WL 2942348 (D. Nev. Sep. 29, 2026), is the second recent Nevada A.I. decision involving use-of-force expert Tom Melton. At issue was discovery of the expert’s 40-page ChatGPT history log.

The first was discussed in Opposing Expert’s Allegedly Inaccurate Description of Use of A.I. Did Not Justify Modification of Scheduling Order on Facts Presented (Sep. 15, 2026), citing Martin v. U.S.A., 2026 WL 2582963 (D. Nev. Sep. 1, 2026).

Villanueva was before the court on objections to the report of a U.S. Magistrate Judge. Plaintiff’s claim arose out of his incarceration by the County.

BELATED DISCLOSURE OF EXPERT’S USE OF A.I.

Plaintiff’s expert testified that he relied on ChatGPT to research and draft his expert report. Id. at *2. However: “Prior to the deposition, Plaintiff did not disclose Melton’s usage of artificial intelligence in response to Defendant’s prior subpoena for ‘all data, documents, and other information considered by expert witness in forming the expert’s opinions.’” Id.

After the deposition, “Defendants requested Plaintiff’s production of the ChatGPT history log….” Id. Plaintiff refused “and subsequently designated [the history log] as protected by ‘Attorney/Expert Work Product; Trial Preparation: Materials’ within their discovery privilege log.” Id. Defendants moved to compel.

THE MAGISTRATE JUDGE’S RULING

The Magistrate Judge granted the motion to compel, holding “that the history log should have been disclosed under Federal Rules of Civil Procedure 26(a)(2)(B)(ii) and (iii) ‘because Melton considered it in forming and supporting his opinions.’” Further, it was not protected as work product because it was not a draft report nor was it created by the expert. Finally, the Magistrate Judge held that the log was not protected by the attorney-client privilege and was relied on by the expert in forming his opinion.

THE PRINCIPLES GOVERNING EXPERT DISCOVERY

Upon objection, the District Judge agreed in part and disagreed in part. The court’s decision explained the principles governing expert discovery:

Expert witnesses may also invoke discovery privileges related to their draft reports. In their expert report, expert witnesses are required to include “the facts or data considered by the witness in forming” their opinion. Fed. R. Civ. P. 26(a)(2)(B)(ii). Rule 26(a)(2)(B)(ii) was amended in 2010 to limit disclosures to “facts or data considered,” instead of previous language on “data or other information” that was interpreted by many courts as requiring the disclosure of “all attorney-expert communications and draft reports.” Fed. R. Civ. P. 26 advisory committee’s note to 2010 amendment. The rules distinguish between the required disclosure of facts or data considered, and an expert’s drafts of their report, which are protected from discovery. “Rules 26(b)(3)(A) and (B) protect drafts of any [expert] report or disclosure required under Rule 26(a)(2).” Fed. R. Civ. P. 26(b)(4)(B).

Id. at *3.

APPLICATION OF THE PRINCIPLES

Components of the A.I. History Log That Must Be Disclosed

The Villanueva court ordered disclosure of those parts of the ChatGPT history log that constitute “facts or data” within the meaning of Rule 26(a)(2)(B)(ii). In the Ninth Circuit, this encompasses “any material considered by the expert, from whatever source, that contains factual ingredients,” and includes material “considered,” not just material “relied upon by the expert.” Id.

One category of chat history sought was “material [that] asks for a case summary of certain controlling court cases on excessive force, in other words legal research conducted by the expert….” Id. at *2.

Looking to precedent, the court wrote: “The legal case summaries generated by ChatGPT constitute ‘facts or data’ that Melton considered and must be disclosed under Rule 26.” Id. at *3. In short:

Plaintiff explains that Melton used the AI software to ask “for a case summary of certain controlling court cases on excessive force.” … Because Melton generated case summaries and “consulted” them to reach his expert opinion on excessive force, the ChatGPT-generated case summaries constitute “facts or data” that are subject to disclosure under Rule 26.

Id. at *3. The court added: “Because AI-generated case summaries are “facts or data,” disclosure under Rule 26 requires the disclosure of all case summary responses supplied by ChatGPT. As Defendant correctly contends, the history log is the only place where Defendant can review alleged ‘hallucinated case citations…quotes…and references to broad legal standards’ that Melton relied upon, or at least considered, to reach his opinion.” Id. (emphasis added).

Components of the A.I. History Log That Are Protected From Disclosure

The court found it was clear error to order production of the “entirety” of Mr. Melton’s ChatGPT history. It held that “Melton’s: 1) questions and prompts to ChatGPT, 2) use of the program to create a deposition outline, and 3) creation of a draft report are all protected from disclosure,” addressing each individually. Id. at *3.

First, Mr. Melton’s questions and prompts were protected because they were neither facts nor data considered by him. “The Court finds that a question is not something that an expert can ‘consider’ or ‘rely’ upon as ‘facts or data’ in reaching their opinion, just as an expert would not ‘consider’ or ‘rely’ upon their online research queries.” Id. at *4. It wrote that, regardless of whether the prompts were viewed as part of an expert’s draft report or trial preparation materials, they were core work product: “The Court finds Plaintiff’s argument persuasive that Melton’s questions and prompts to ChatGPT reveal his ‘mental impressions or strategy…as to what issues he found of particular importance’ and Melton’s theory of the case.”

Second, the ChatGPT log used to predict likely questions was protected. “Unlike case summaries, articles, or print materials relied upon by an expert to reach their opinion, the Court finds that a deposition outline is not ‘facts or data’ within the meaning of Rule 26. Instead of using ChatGPT’s outputs to inform his expert opinion, Melton used ChatGPT’s responses to structure his preparation for his deposition and trial. Plaintiffs should not have to disclose the portion of the history log related to deposition preparation.” Id. at *4. Whether viewed as protected trial preparation materials or, alternatively, Mr. Melton’s “mental impressions, conclusions, opinions, or legal theories,” the logs were protected because disclosure “would reveal Melton’s focus, priorities, or areas of concern to address in the deposition and at trial.” Id.

Third, “the portions of the history log related to Melton’s creation of a draft report” were not discoverable. Id. at *4.

The Court finds that Melton’s prompts to create a draft report on “certain issues, such as qualifications of the opposing expert or analysis of the opposing expert’s report”… is squarely within the protections of an expert’s draft reports. Rule 26(b)(4)(B) mandates the protection of expert drafts “regardless of the form in which the draft is recorded.” Melton testified to sometimes copying and pasting the outputs of ChatGPT into his own report…. It appears that Melton’s usage of the ChatGPT responses was to assist in his writing and analysis of his expert report, rather than to create “facts or data” to consider in forming his opinion.

Id. (emphasis added). The court rejected the defense argument that these were not Mr. Melton’s “own generation of a report….” Id.

Melton testified that some of this information was directly incorporated into his own written report. Even if the ChatGPT output is not a draft report protected under Rule 26(b)(4)(B), it would still be mental impressions protected from disclosure under Rule 26(b)(3)(b). As addressed above, the ChatGPT responses would provide direct insight into Melton’s “mental impressions” and “legal theories” that he considered relevant in a report.

Miscellaneous Objections

Plaintiff asserted that it was clear error for the Magistrate Judge to find that privilege was waived by Mr. Melton’s use of ChatGPT. However, the court wrote that “Judge Albregts did not reach a decision on whether the log was privileged, and if so, if any such privilege was waived by Melton’s disclosure of information to ChatGPT.” It added:

Because ChatGPT is not a person, but was an application used in this instance as a research, drafting, and editing program, the Court finds that any privileged information is not waived by Melton’s use of the program. The Court also notes that many of Defendant’s concerns about Melton’s methodology and reliance on ChatGPT are appropriate to address at trial as matters of impeachment.

Id. at *5 (emphasis added).

Plaintiff also objected that it was clear error for the Magistrate Judge not to conduct in camera review. The District Judge, having conducted an in camera review, disagreed.

For more on in camera review, please visit: