June 2026


Client Alert - AI & Litigation Risk - Part 2

Courts Are Regulating Your AI Tools. Are You Ready?



 
 

Courts are no longer asking whether AI belongs in litigation. They are regulating it in specific terms, with enforceable consequences. In the past three months alone, federal courts have banned certain AI platforms from litigation, required parties to disclose which AI tools were used to process confidential documents, and found AI-generated work product not reviewed by attorneys improper. If your legal team has not audited its AI practices, it needs to do so now.


eDiscovery: Courts Draw the Line


Two recent protective order rulings set the new floor for any AI tool used in discovery.


In Jeffries v. Harcros Chemicals, Inc., 2026 WL 820218 (D. Kan., Mar. 25, 2026), a Kansas federal court amended its protective order to prohibit open AI platforms from document review entirely. Only closed AI tools are permitted, a ruling driven by data privacy obligations, including GDPR, and concern for the chilling effect on document production for fear files will be fed to open AI systems.


Morgan v. V2X, Inc., 2026 WL 864223 (D. Colo., Mar. 30, 2026) went further, establishing a three-part test that any AI tool must satisfy before it can touch confidential discovery materials:


  • Training Ban. The AI provider must be contractually prohibited from using your inputs to train its models.
  • Strict Confidentiality. The provider cannot share your inputs with third parties beyond what service delivery requires.
  • Right to Delete. You must be able to require deletion of all submitted information.


The court also ordered the party to disclose which AI platform it had used to process confidential materials.


Free-tier, open AI platforms fail this test. An enterprise agreement with these data protection provisions is the minimum. Build these requirements into your protective order template now. Courts in both cases made clear they will grant motions to amend orders that ignore AI.


Your Experts Are Not Shielded


Conservation Law Foundation v. Shell Oil Co., Case No. 3:21-cv-00933 (VDO) (docket entry only) (D. Conn., May 18, 2026; stayed June 3, 2026) pushed the AI perimeter into expert witness territory. Plaintiff’s expert used an AI tool to filter and categorize a large document production. The court ordered production of the AI prompts, finding that how an expert narrows the document universe is part of her methodology, and methodology is discoverable under Rule 26. A Rule 29 stipulation covering “notes and drafts” was not specific enough to cover AI prompts.


The ruling is stayed pending a Rule 72(a) objection, and the district court’s ruling on the objection will be the more consequential decision for practitioners. But the risk is now clear. An expert’s AI interactions may become part of the expert file. Address this explicitly in every Rule 29 agreement and Rule 26(f) plan.


Ethics Rules: Obligations Your Lawyers May Not Know They Have


White v. Walmart, Inc., 2026 WL 1002132 (S.D. Ind., Apr. 14, 2026) gave the clearest judicial statement yet on what it looks like when AI replaces attorney judgment. Plaintiff’s counsel uploaded defendant’s discovery responses to an AI tool, asked it to identify deficiencies, and forwarded the output, unreviewed, to opposing counsel and the court. The AI flagged every single interrogatory response as deficient. The court was unimpressed. It held that exclusive reliance on AI output violates Rule 37’s meet-and-confer obligation and called the approach a “perilous shortcut.”


The rule established by this case is simple: AI produces the draft. The attorney has to make the call. Every item. Every time. This is not limited to discovery. It applies to everything that leaves your lawyers’ hands.


The NYC Bar’s Formal Opinion 2025-6 (Dec. 2025) addresses a risk most clients overlook: AI recording tools on attorney-client calls. Your outside counsel cannot use these tools on a call with you without your informed consent regardless of one-party consent law. More importantly, your own use of these tools on calls with counsel creates confidentiality and discovery risk. If you are recording calls with outside counsel using any AI notetaker, those recordings and summaries may be discoverable. Your retainer agreements should address this, and your outside counsel should be asking.


Oregon State Bar Formal Opinion 2026-208 addresses similar concerns for use of autonomous AI agents for client intake. They are permissible but only with active attorney oversight. Any AI agent must not inadvertently create attorney-client relationships, make unauthorized representations, or compromise prospective client confidentiality.



California: The Shift from Guidance to Binding Rules


California is laying the groundwork to become the first U.S. jurisdiction to make AI ethics violations disciplinable offenses. The California State Bar’s Standing Committee on Professional Responsibility and Conduct is reviewing proposed amendments to six Rules of Professional Conduct (covering competence, confidentiality, candor, and supervision) that would convert current advisory guidance into binding rules carrying potential discipline. California’s Supreme Court must adopt them before they take effect. But if California moves, other states will follow.

 

What to Do Now


  • Audit every AI tool used by your litigation team and outside counsel. Any tool that touches confidential discovery materials must satisfy the Morgan three-part test (training ban, strict confidentiality, right to delete). Free-tier tools do not.
  • Add AI governance language to your standard protective order template covering the Morgan test, platform name disclosure, and a GDPR data processing clause for any matter involving EU personal data.
  • When you retain an expert, ask whether they use AI in their work. If yes, establish a preservation protocol for AI prompt records at the start of the engagement and address expert AI use explicitly in your Rule 29 agreement.
  • Require that every AI-generated output be independently reviewed by an attorney before it goes anywhere.
  • Implement a policy on AI meeting recorders for calls with counsel. Require informed consent and enterprise-grade tools. Brief your executive team that their AI notetaking apps may create discovery exposure.
  • Update your litigation hold notices to prohibit custodians from processing hold materials through consumer AI tools.
  • Privilege logs for AI materials must identify the platform used, whether it satisfies Morgan, who directed the interaction, and what confidentiality protections were in place.


We are available to help assess your organization’s exposure, update your internal policies, and structure your AI use to preserve the protections that matter. For any questions, please contact Joshua Leader at (212) 486-2400 | jleader@leaderberkon.com.


  

© 2026 Leader Berkon Colao & Silverstein LLP | Attorney Advertising | This alert is for informational purposes only and does not constitute legal advice.

 



New York | Philadelphia | Los Angeles | Dallas

 
 

leaderberkon.com