Court Orders Disclosure Of Expert Witness’s AI Prompts: What Litigators Need To Know

On May 18, 2026, Magistrate Judge Thomas O. Farrish of the US District Court for the District of Connecticut ordered the plaintiff in Conservation Law Foundation, Inc. v. Shell Oil Company, et al. (Case No. 3:21-cv-00933, D. Conn.), to produce the generative AI prompts that its expert witness, Dr. Naomi Oreskes, used in preparing her expert report. (ECF No. 970). This appears to be the first federal court decision requiring an expert witness to disclose AI prompts as part of discoverable methodology for the expert’s opinions. The district court has stayed the order pending resolution of plaintiff’s objection.

Background

The plaintiff, Conservation Law Foundation (CLF), retained Dr. Oreskes as an expert on the history of climate change science. In preparing her expert report, Dr. Oreskes and her research assistant, Dr. Kaurov, used a commercially available generative AI tool to filter and identify potentially relevant documents from the defendants’ document productions. Dr. Oreskes disclosed her use of AI in her expert report, served on May 8, 2025, identifying certain search terms and frequencies. Defendants subsequently sought additional information about the AI prompts and system outputs through Dr. Oreskes’ deposition on September 9, 2025, and various informal requests.

After extended meet-and-confer efforts failed to resolve the dispute, Defendants filed a motion to compel. On May 18, 2026, Magistrate Judge Farrish granted the motion, finding that “[a]n expert witness’s methodology is fair ground for discovery,” and that “the process by which Dr. Oreskes culled down the defendants’ document production into a subset to be worked with is an aspect of that methodology.” The court rejected CLF’s argument that a stipulation protecting expert “notes, drafts, or communications” shielded the prompts, holding that the agreement was not “quite clear” enough to cover AI prompts specifically. Finally, the court found that defendants had an “evidence-backed reason” for doubting CLF’s representation that no additional materials existed, because the expert’s research assistant had referenced “prompt[s]” in his own declaration.

Pending Objection

On June 2, 2026, CLF filed an objection to the Magistrate Judge’s order under 28 U.S.C.
§ 636(b)(1)(A) and Fed. R. Civ. P. 72(a), arguing the order is “clearly erroneous or contrary to law” on multiple grounds, including: (1) the court applied the broad Rule 26(b) relevance standard rather than the narrower expert disclosure requirements of Rule 26(a)(2)(B)(ii); (2) no formal Rule 33 or Rule 34 requests were ever served, so Rule 37 cannot enforce compliance; (3) the Rule 29 stipulation expressly protects such iterative research materials; and (4) the motion was untimely and overlaps with defendants’ pending Daubert challenge. CLF further argues that the sole precedent cited, Macchia v. ADP, 711 F. Supp. 3d 162 (E.D.N.Y. 2024), involved fact, not expert, discovery. The district court judge has stayed the order pending resolution of CLF’s objection. The matter remains pending.

Broader Implications for Expert Discovery

This order, if upheld, could meaningfully expand the scope of expert discovery. By treating AI prompts as discoverable methodology, rather than as the functional equivalent of research assistant communications or keyword searches, the court has signaled that any AI-assisted step in the report preparation process may be subject to disclosure, regardless of whether the expert actually “considered” the AI outputs in forming opinions or used the tool only for preliminary document culling. The practical effect is to place AI prompts in the same category as dataset selection criteria or statistical model parameters—aspects of methodology that have long been fair game, rather than treating them as protected preliminary research. The precedential basis for this expansion is thin.

The ruling also raises the prospect of increasingly granular discovery into AI-assisted workflows. If prompts are discoverable, parties may face broad demands for prompt histories, iterative refinements, system instructions, and chat logs, even when such materials were not preserved in the ordinary course. This could create significant burdens, particularly where AI tools are used interactively and iteratively, and courts will need to develop frameworks for distinguishing between AI interactions that genuinely inform an expert’s methodology, and those that are more analogous to preliminary, discarded research—a line that existing precedent on “facts or data considered” under Rule 26(a)(2)(B)(ii) does not clearly draw.

Practical Recommendations

In light of this development, outside counsel, in-house teams, and experts should consider the following proactive steps.

• Establish internal protocols for expert AI use. Develop clear guidelines governing when and how retained experts may use generative AI tools in connection with litigation. Protocols should address which AI tools are permissible, how prompts should be formulated, and what records (if any) should be preserved. Treat AI prompts as part of the expert’s methodology from the outset.
• Address AI use in expert-retention agreements. Include provisions in expert-retention letters or engagement agreements that address the expert’s use of AI, preservation obligations, and disclosure expectations. Consider specifying whether the expert is authorized to use AI, and if so, what documentation requirements apply.
• Raise AI issues early in litigation. At Rule 26(f) conferences or in proposed discovery plans, seek stipulations or court orders that explicitly define the scope of AI-related discovery—including whether prompts, system instructions, and iterative query logs are within or outside the scope of production. Propose protective-order language addressing the treatment of AI prompts and logs, particularly where such materials may reveal proprietary methodologies or attorney work product.
• Conduct AI-use audits before expert reports are finalized. Before expert reports are served, review whether the expert used AI tools during the engagement and assess whether the use and any outputs should be disclosed proactively to minimize the risk of later disputes. Address AI explicitly in Rule 29 agreements, as generic references to “notes” or “communications” will not suffice after this ruling.
• Monitor developments. This is an evolving area with no established precedent. Practitioners should monitor the resolution of the pending objection and any future decisions addressing the discoverability of AI prompts in the expert-witness context.

Visit us at mayerbrown.com

Mayer Brown is a global services provider comprising associated legal practices that are separate entities, including Mayer Brown LLP (Illinois, USA), Mayer Brown International LLP (England & Wales), Mayer Brown (a Hong Kong partnership) and Tauil & Chequer Advogados (a Brazilian law partnership) and non-legal service providers, which provide consultancy services (collectively, the "Mayer Brown Practices"). The Mayer Brown Practices are established in various jurisdictions and may be a legal person or a partnership. PK Wong & Nair LLC ("PKWN") is the constituent Singapore law practice of our licensed joint law venture in Singapore, Mayer Brown PK Wong & Nair Pte. Ltd. Details of the individual Mayer Brown Practices and PKWN can be found in the Legal Notices section of our website. "Mayer Brown" and the Mayer Brown logo are the trademarks of Mayer Brown.

© Copyright 2026. The Mayer Brown Practices. All rights reserved.

This Mayer Brown article provides information and comments on legal issues and developments of interest. The foregoing is not a comprehensive treatment of the subject matter covered and is not intended to provide legal advice. Readers should seek specific legal advice before taking any action with respect to the matters discussed herein.