Morgan v. V2X, Inc. is an early court order on AI and discovery confidentiality. In March 2026, a Colorado federal magistrate judge held that work-product protection can cover a pro se litigant's AI use, but that the name of the AI tool used with confidential information is not protected, and amended the protective order to set contractual conditions for any AI platform that receives confidential discovery materials. Every attorney handling confidential documents should understand this order because the court said it will, for now, bar most mainstream low-cost and free AI tools (standard ChatGPT, Claude and Gemini) from receiving protected material. It is one magistrate judge's order in one case, not appellate authority. The ruling sits alongside Mata v. Avianca as one of the two foundational AI sanctions cases practitioners reference: Mata addressed hallucinated citations in filed briefs; Morgan addresses confidentiality obligations when AI processes protected discovery. Where Mata turned on FRCP Rule 11 certification obligations, Morgan turns on protective order compliance under Rule 26.


Background

Morgan v. V2X started as a routine employment discrimination case in the District of Colorado, filed in June 2025 as Case No. 1:25-cv-01991-SKC-MDB (cited as Morgan v. V2X, Inc., No. 1:25-cv-01991 (D. Colo. Mar. 30, 2026)). The plaintiff, proceeding pro se (without an attorney), received confidential discovery materials from the defendant, V2X, Inc., under a standard protective order. These materials contained the kind of sensitive information that protective orders exist to safeguard.

The order records that both parties were using AI in their litigation work and disagreed about how it could be used with Confidential Information; it makes no finding that the plaintiff uploaded protected material to any particular tool. AI use in discovery is increasingly common: parties use ChatGPT, Claude, or other AI assistants to help understand complex legal documents, draft responses, or identify relevant information. For a pro se litigant without legal training, the appeal of AI assistance is obvious.

V2X moved to amend the protective order to address AI and asked the court to compel the plaintiff to identify his AI tool. The defendant's concern was specific: consumer AI tools process user inputs on remote servers, and some use that data to train their models. Uploading confidential discovery materials to these platforms could expose protected information to third parties or embed it in AI training data permanently.

Morgan v. V2X, Inc.
No. 1:25-cv-01991 (D. Colo. Mar. 30, 2026)
Court
U.S. District Court, District of Colorado
Date
2026-03-30
Category
AI Privacy / Confidentiality
Sanctions
None
AI Case Law: Updated April 2026 · Primary source: Order, ECF 65 (CourtListener)

What Happened

Magistrate Judge Maritza Dominguez Braswell faced a question about AI and protective orders: does uploading confidential discovery materials to AI tools violate a protective order, and if so, what safeguards are needed? The court took the question seriously, issuing a detailed ruling on March 30, 2026, that addressed both the work product implications and the practical risks.

The court first held that a party's use of AI in litigation may reflect mental impressions and legal theories protected as work product under Rule 26(b)(3). What you ask an AI, how you frame your questions, and what documents you select for analysis can reveal your litigation strategy. That's protected.

But the court also found that the identity of the AI tool used is not protected. Opposing parties and the court have a right to know which AI platforms are processing confidential materials. This distinction matters: your strategy is protected, but your tool choices aren't.


The Ruling

The court amended the protective order, using its own language rather than either party's, to set conditions for any use of AI tools with confidential discovery materials. No party or authorized recipient may input, upload or submit CONFIDENTIAL information into a modern AI platform unless the provider is contractually bound in three ways. First, it must be prohibited from storing or using inputs to train or improve its model. Second, it must be prohibited from disclosing inputs to any third party, except where essential to delivering the service, and then the third party must be bound by obligations no less protective. Third, it must contractually allow the removal and deletion of all CONFIDENTIAL information on request. A party relying on an AI tool must also keep written documentation of these contractual protections.

The court said these conditions will, for now, bar most mainstream low-cost and free AI tools (standard ChatGPT, Claude and Gemini) for Confidential Information. Even paid versions of consumer AI tools don't necessarily provide contractual guarantees about data handling. Enterprise AI platforms with specific data processing agreements are the likelier fit. AI may still be used for work that doesn't involve uploading Confidential Information. The court also rejected the plaintiff's proposed "closed-circuit environment" carve-out, reasoning that it addressed cybersecurity rather than AI-specific risk.

The court also ordered the plaintiff to disclose, within 10 days, the name of any AI platform used with information designated CONFIDENTIAL. This disclosure requirement creates accountability: parties can't quietly upload protected documents to AI tools and hope nobody notices. The order tries to balance AI's practical benefits with the confidentiality obligations that discovery depends on. It imposed no sanction or fine, and it is a magistrate judge's order, not appellate authority.

Outcome: The court held that AI use by litigants is protected by the work product doctrine under Rule 26(b)(3), but ordered the plaintiff to disclose the identity of any AI tools used with confidential materials. The court amended the protective order to impose three requirements: AI providers must be contractually barred from training on inputs, cannot disclose inputs to third parties, and must allow deletion of all confidential data.

Why This Case Matters

This order is worth studying before your next protective-order negotiation. It binds only the parties in that case, but it shows how a court can handle AI and confidential discovery materials. Some attorneys use consumer AI freely. Others avoid it entirely. This order sets a middle path: you can use AI with confidential information, but only through platforms that meet the court's three contractual requirements, and with written documentation.

The practical impact depends on your protective orders, but the audit is worth doing now. Firms handling cases with protective orders should review which AI tools their attorneys and staff use. Under this order's terms, consumer-grade ChatGPT, free Claude and Google Gemini without enterprise agreements would not qualify for confidential materials. Firms need enterprise AI platforms with proper data processing agreements or risk violating protective orders that adopt similar terms.

We have not seen evidence that other courts have adopted this order's language. Parties and courts drafting AI-aware protective orders may borrow from it, since the three requirements (no storing or training, no third-party disclosure beyond what the service needs, deletion rights) are specific and easy to copy. Treat it as one court's approach, not as an established national standard.


Lessons for Attorneys

Every managing partner needs to issue clear guidance on AI tool use with confidential materials. The Morgan v. V2X order gives you a model: platforms with contractual data protections only, for confidential material. No consumer AI tools for anything covered by a protective order. Train your attorneys and staff on the distinction, because a single upload to the wrong platform can violate a court order.

For litigators specifically: review every active case with a protective order and assess whether any AI tools have been used with protected materials. If they have, determine whether the tools meet the court's three requirements. If they don't, you may need to disclose the issue to the court and opposing counsel. Self-reporting is painful, but it's better than having it discovered later.

For solo practitioners and small firms without enterprise AI budgets: this ruling creates a cost barrier. Enterprise AI platforms with proper data agreements are more expensive than consumer tools. But the alternative is risking sanctions for violating protective orders. Factor AI platform costs into your litigation budget the same way you factor in Westlaw or document review tools.


The Bottom Line

Morgan v. V2X conditions the use of AI with confidential discovery materials on a three-part contractual test: no storing or training on inputs, no third-party disclosure beyond what the service needs, and deletion rights, with written documentation. The court said it will, for now, bar most mainstream low-cost and free AI tools. Every firm handling protected materials needs platforms that meet those terms or needs to stop using AI for those documents.

AI-Assisted Research. This piece was researched and written with AI assistance, reviewed and edited by Manu Ayala. For deeper takes and the perspective behind the research, follow me on LinkedIn or email me directly.

Price sources, checked 22 October 2026: Gavel. Each figure above is a list price from the vendor's own pricing page, or arithmetic on one. Plans, seat minimums and promotions change; confirm on the vendor's page before you buy.