Use of AI Does Not Eliminate All Expectations of Privacy, Says Court: EDiscovery Case Law

Key Takeaways

• AI use does not waive work‑product protection under Rule 26(b)(3).
• Plaintiff must disclose any AI tool used with confidential data.
• Court crafted balanced protective order requiring contractual safeguards on AI providers.
• Pro se litigants may use AI if no confidential inputs are uploaded.

Pulse Analysis

The Morgan v. V2X decision arrives at a pivotal moment as law firms and corporate legal departments increasingly rely on large language models for document review and analysis. By distinguishing civil work‑product doctrine from criminal precedent, the court reaffirmed that a pro se litigant’s AI‑generated notes and drafts are covered by Rule 26(b)(3). This protects strategic thinking and mental impressions, even when the underlying technology processes data on external servers, underscoring that privacy expectations persist despite the cloud‑based nature of modern AI.

Judge Dominguez Braswell’s protective‑order template sets a practical standard for AI governance in eDiscovery. The order mandates that any AI platform handling confidential information must contractually prohibit data retention for model training and limit third‑party disclosures. It also requires parties to retain written proof of these safeguards and to delete inputs on request. By rejecting overly restrictive and overly permissive proposals, the court offers a middle ground that balances risk mitigation with the efficiency gains AI can deliver, giving litigants a clear compliance roadmap.

The broader impact extends beyond this single case, signaling to the industry that AI tools will be subject to the same confidentiality regimes as traditional software. Litigation support providers must now embed contractual clauses that meet the court’s criteria, and eDiscovery teams will need to audit their AI workflows for inadvertent data uploads. As courts across the country grapple with similar issues, this ruling provides a persuasive template that could shape future protective‑order drafting, influencing how AI is integrated into the discovery process nationwide.