Order Prohibiting Upload of Confidential Discovery Documents to Artificial Intelligence (“AI”)

[EDRM Editor’s Note: The opinions and positions are those of Michael Berman.]

“[T]he Court’s March 18, 2025 Fed. R. Civ. P. 26(c) protective order … is hereby modified to reflect that any documents marked confidential shall not be uploaded onto any AI platform.” Warner v. Gilbarco, Inc., 2025 WL 3047881 (E.D. Mich. Oct. 30, 2025)(emphasis in original).

Apparently, the concern was that, if confidential documents were uploaded to AI, confidentiality would be lost. As such, the issue presents a yellow caution light for many protective orders. This may be an issue to discuss in a Rule 26(f) conference of the parties.

In Warner, the plaintiff alleged discrimination based on race. The order was issued in response to pro se plaintiff’s motion to compel defendants to produce, among other things, “comparator personnel files, HR investigation files, training manuals, termination memos, internal communications, and related materials….”

The court wrote that:

Defendants’ reluctance to produce certain records – e.g., those related to Shawn McClellan (in response to RFP No. 4) or to Paul Blaser (in response to RTP No. 5), or to the three people involved in her termination (in response to Int. No. 7 & RFP No. 3) – relates to concerns that the information will be uploaded to an AI platform. [emphasis added].
Warner v. Gilbarco, Inc., 2025 WL 3047881 (E.D. Mich. Oct. 30, 2025).

The court amended the protective order “[f]or reasons set forth on the record” of the prior hearing, but not explained in the written decision. As of this date, a transcript of the hearing is not available online.

Much of the interest in AI has centered on hallucinations, which were also present in this case, id. at *3, and discoverability of prompts or training materials.

Here, the issue was how confidential materials produced in discovery can be used without exposing them to loss of confidentiality. One may guess that a pro se plaintiff would be using an inexpensive or free AI program that does not protect confidentiality.

Here, the issue was how confidential materials produced in discovery can be used without exposing them to loss of confidentiality.
Michael Berman, E-Discovery LLC.

Here, the court’s order referred to “any” AI programs. On the facts before the court, that broad order may have been – and very likely was – entirely justified.

However, many AI programs, including for example, Lexis, Westlaw, and DISCO, to give only very few examples, protect confidentiality of documents uploaded to their AI platforms.

It would seem that a broad order such as that in Warner would be unduly restrictive in that context. Discovery issues related to electronically stored information (“ESI”) are always fact sensitive.

Assisted by GAI and LLM Technologies per EDRM GAI and LLM Policy.

Author

Michael D. Berman

Mike is the owner of E-Discovery, LLC, and of counsel at Rifkin Weiner Livingston LLC, in Baltimore. He concentrates on commercial litigation and offers mediation services. He was the primary editor of Electronically Stored Information in Maryland Courts (Md. State Bar Ass’n. 2020), and he co-edited M. Berman, C. Barton, and P. Grimm, eds., Managing E-Discovery and ESI: From Pre-Litigation Through Trial (ABA 2011), and J. Baron, R. Losey, and M. Berman, eds., Perspectives on Predictive Coding (ABA 2016).

Mike has litigated a number of cases in the trial and appellate courts in Maryland. He is an Adjunct Professor at the University of Baltimore School of Law where he co-teaches a three-credit discovery workshop that focuses on e-discovery. He has lectured at the Maryland Judicial College and he chaired the Bar committee that drafted the proposed ESI Principles for the District of Maryland. He is a past: co-chair of the Federal District Court Committee of the Maryland State and Federal Bar Associations; chair of the Litigation Section Council, Maryland State Bar Association; and, co-chair of the American Bar Association Litigation Section Book Publishing Board. He graduated from the University of Maryland School of Law and is also an Army veteran. He is admitted to the Maryland bar. The opinions expressed in this blog are not necessarily those of Rifkin Weiner Livingston LLC.

View all posts