In Morgan v. V2X, Inc., an employment discrimination case brought by a pro se plaintiff in the U.S. District Court in Colorado, the court addressed a discovery dispute arising from the parties’ use of AI in litigation. The conflict emerged after the defendant sought to amend the protective order to regulate AI use and to compel disclosure of the specific AI tool the plaintiff had used in litigation, citing concerns that confidential information may have been uploaded to third-party platforms.
The court held that Federal Rule of Civil Procedure 26(b)(3) extends work-product protection to a pro se litigant’s use of AI, including AI-assisted litigation materials that may reflect mental impressions or strategy; however, the party still needs a factual basis showing the tool itself would reveal mental impressions. The court did draw a distinction between protected substantive outputs and non-protected facts about the process itself. It ordered the plaintiff to disclose the identity of any AI tool used, finding that the plaintiff failed to show that such disclosure would reveal litigation strategy. At the same time, the court rejected the assertion that using AI tools automatically waives work-product protection, reasoning that disclosure to an AI platform is not equivalent to disclosure to an adversary and, by itself, does not substantially increase the likelihood that protected material will reach opposing counsel.
Importantly, rather than resolving these disputes solely through privilege doctrine, the court imposed forward-looking safeguards by amending the protective order to strictly limit the use of AI with confidential information. The order effectively bars parties from inputting protected material into most standard and free versions of mainstream AI tools, including ChatGPT, Claude, and Gemini, unless the provider offers specific contractual guarantees regarding data use, retention, and deletion. The decision underscores an emerging trend in which courts treat certain AI-related facts as discoverable, preserve core work-product protections notwithstanding AI use, and rely on protective orders as the primary mechanism to manage confidentiality risks in an evolving technological landscape.
