In a May 5, 2026, blog post, we examined Morgan v. V2X, Inc. and the court’s analysis of whether the work product doctrine protects a pro se litigant’s use of AI. Magistrate Judge Maritza Dominguez Braswell held that Federal Rule of Civil Procedure 26(b)(3) applies to pro se litigants and extends to their use of AI tools — meaning that AI-assisted litigation preparation may qualify as protected work product. She further held, however, that work product protection is not unlimited: the court ordered Morgan to disclose the name of the AI platform he used to process “confidential information” because he failed to demonstrate that doing so would reveal his mental impressions or case strategy.
In this post, we examine the second issue addressed by the court: to what extent should a protective order expressly restrict the use of AI? Both parties used AI in connection with confidential discovery materials reviewed and exchanged in discovery — and each proposed competing protective order language to address the concern. Accordingly, the court was called upon to craft a workable standard. The result is a provision that may carry practical implications for litigants, particularly those relying on mainstream, low-to-no-cost AI platforms.
The Operative Protective Order and Competing Proposals
The operative protective order in Morgan v. V2X, Inc, prohibited disclosure of “confidential information” except under specifically enumerated circumstances. As Judge Dominguez Braswell noted, the existing order arguably already covered AI use, but the parties agreed that express AI-specific language was necessary to provide clarity.
The parties embraced competing visions of how the protective order should address AI use. V2X proposed a framework that would prohibit authorized recipients of confidential information from using any application or software that might transfer confidential information to another provider — unless the receiving party confirmed, through due diligence, that the provider’s security and privacy controls would allow compliance with the protective order. V2X’s proposed order would also mandate that any such system allow for removal or deletion of confidential information from the system. V2X’s proposal would also have expressly prohibited the use of confidential information to train any AI tool and identified specific platforms by name.
Morgan, on the other hand, proposed a more principles-based approach. Morgan’s proposal would have allowed the use of AI tools so long as they operated within a “secure, closed-circuit environment,” and would have barred parties from uploading confidential information to platforms whose terms of service allowed the provider to use that data to train large language models, machine learning algorithms, or for internal human-in-the-loop review.
The Court’s Provision for AI Use with Confidential Information
The court rejected both parties’ submissions and crafted its own provision:
No party or authorized recipient may input, upload, or submit CONFIDENTIAL Information into any modern artificial intelligence platform, including any generative, analytical, or large language model-based tool (‘AI’), unless the AI provider is contractually prohibited from: (1) storing or using inputs to train or improve its model; and (2) disclosing inputs to any third party except where such disclosure is essential to facilitating delivery of the service. Where disclosure to a third party is essential to service delivery, any such third party shall be bound by obligations no less protective than those required by this Order. In addition, the AI provider must contractually afford the party or authorized recipient the ability to remove or delete all CONFIDENTIAL information upon request. A party intending to use AI that it contends meets these requirements must retain written documentation of these contractual protections.
Noting the impact of this standard “[t]he Court recognize[d] that practically speaking, and in light of the current state of AI, this provision will (at least for now) bar the parties from using most, if not all, mainstream low-to-no-cost AI to process Confidential Information.” (emphasis added).
In other words, free or low-cost AI platforms — the tools most accessible to individual litigants and smaller firms — may be unlikely to satisfy the court’s contractual safeguard requirements. Enterprise-tier AI accounts that meet the court’s standards, such as those offered through organizational procurement processes, may be cost prohibitive to a pro se litigant or a litigant of limited means. The court acknowledged this tension directly, noting that as large firms invest in enterprise-grade AI and “make their use of AI more secure, efficient, effective, and powerful,” the question of how pro se litigants and under-resourced parties will keep up is a “growing problem in the age of AI.”
To address this concern, the court took two steps. First, it cautioned the parties against over-designating information as confidential, recognizing that the breadth of the confidentiality designation directly affects the scope of the AI restriction. Second, it clarified that the order does not restrict all AI use — only the use of AI to process confidential information. As the court stated, “Modern AI tools may be used in many ways that do not involve uploading Confidential Information, and nothing in this particular Order restricts those uses.”
Takeaways
The Morgan v. V2X decision carries practical lessons for parties:
- Courts are scrutinizing AI use in litigation, so parties may encounter questions about the use of AI from opposing counsel and the bench.
- Uploading confidential discovery materials to free or low-cost platforms is not without risk and may not currently meet Morgan’s contractual safeguards or other standards set forth in operative protective orders.
- While courts are beginning to recognize that AI-assisted litigation preparation may qualify as protected work product, that protection has limits. Simply naming the tool used may not shield parties from disclosure if their opponent has a legitimate need to know.
- In matters involving protective orders, parties may wish to consider whether express AI-provisions are advisable.
- AI-related eDiscovery case law is a rapidly developing area of law. Cases such as Concord1, Heppner, Warner, and now Morgan, are just the beginning.
1 Concord Music group, Inc. v. Anthropic PBC (finding attorney work product protections may extend to attorney AI prompts.); Concord Music group, Inc. v. Anthropic PBC (ordering production of non-lawyer AI prompts by a party’s employees, subject to usual proportionality and relevancy constraints.)