AI, Work Product, and the Protective Order Problem: What Morgan V. V2X, Inc. Means for Every Litigator

Key Takeaways

• Work product protection extends to AI outputs generated by pro se litigants
• Court rejects waiver argument; third‑party AI does not destroy confidentiality
• Protective order must bar AI training use and require input deletion
• Enterprise‑grade AI may be required; free tools likely non‑compliant
• Undocumented AI platform choices can forfeit privilege and work product

Pulse Analysis

The Morgan v. V2X decision arrives at a moment when generative AI tools are becoming ubiquitous in law firms and corporate legal departments. By anchoring AI‑generated drafts to Rule 26(b)(3), the court affirms that the privilege traditionally reserved for attorney‑crafted work product now covers outputs produced by a litigant acting as his own counsel. This extension reflects the reality that AI can serve as a surrogate mind, enabling pro se parties to formulate legal arguments that would otherwise be inaccessible. The ruling also draws on privacy jurisprudence, emphasizing that merely routing data through a cloud service does not extinguish confidentiality expectations.

Beyond the doctrinal holding, the opinion introduces a protective‑order framework that could become a national template. It obligates any AI provider handling confidential case data to prohibit model training, restrict third‑party disclosures, and delete inputs on demand. While this safeguards sensitive information, it effectively excludes most low‑cost, consumer‑grade AI platforms, pushing litigants toward enterprise‑level solutions that include robust data‑processing agreements. The order also raises technical questions about vector embeddings and whether a “right to delete” can truly erase the mathematical representations that persist in AI models, an issue that will likely spawn further litigation.

Practically, the decision forces litigation teams to inventory every AI tool in use, verify contractual safeguards, and document the specific functions that support case strategy. Counsel must advise clients that prompts and model selections can constitute protected mental impressions, and that failure to establish a factual record may result in loss of privilege. As courts begin to adopt similar language, the compliance burden will fall hardest on smaller firms and pro se parties, making strategic planning around AI an essential component of modern litigation risk management.