Kansas Judge Bars Open-Loop AI Tools for All Discovery Materials

What the court decided

The litigation involves consolidated putative class actions brought by residents living near a Kansas City chemical facility. The residents allege chronic injuries from toxic emissions. The court had already entered a protective order requiring parties to use closed or secure AI tools when handling confidential information.

The defendants then moved to extend those restrictions to all discovery materials produced in the case, whether marked confidential or not. In effect, they sought a blanket ban on uploading any discovery documents into public, open AI tools. Judge Mitchell granted the motion to amend the order.

Everlaw explains the technical distinction behind the ruling. Open loop AI tools operate in public environments and continually ingest all uploaded information to train and develop their underlying machine learning models. Closed loop tools use secure environments and strict contractual safeguards that explicitly prohibit the provider from using the data to train or improve its public models.

How the judge answered the plaintiffs

The plaintiffs argued that the restriction was a disfavored umbrella order that would raise litigation costs and violate their First Amendment rights to disseminate non-confidential discovery materials. According to Everlaw, Judge Mitchell rejected these objections.

She clarified that the order was not an umbrella order, because parties would still screen documents for confidentiality designations. On the First Amendment, the court noted that the order does not stop parties from publicizing non-confidential documents through traditional methods such as websites or court filings. It only controls the interim technology used to process discovery. Judge Mitchell wrote that "A protective order does not offend the First Amendment when it is entered based on a showing of good cause, is limited to the context of pretrial civil discovery, and does not restrict the dissemination of the information if gained from other sources," citing Seattle Times Co. v. Rhinehart, 467 U.S. 20, 37 (1984).

On costs, the court emphasized that the rule applies equally to both sides. Litigants remain free to use generative AI to accelerate document review, draft timelines or summarize evidence, provided they use closed AI tools that meet basic security requirements.

Why open AI tools were seen as a risk

The court found that the defendants showed "good cause" under Federal Rule of Civil Procedure 26(c)(1). It pointed to three main concerns:

• No claw back: Open AI tools use submitted data to keep training and improving their models. Once data is ingested by a public model, it is practically impossible to claw back privileged information or delete it when the lawsuit ends.
• GDPR exposure: Two of the defendants are subject to the European General Data Protection Regulation. The court noted that wholesale submission of discovery data to an open AI tool could expose large amounts of personal data without the consent required by GDPR Article 4(11).
• Critical infrastructure: The defendants operate in the chemical sector, which is designated as critical infrastructure vital to national security. The court validated concerns that exposing confidential data to open AI tools could hand cybercriminals a roadmap for data breaches.

Context and what it means for ediscovery

Everlaw links the decision to Morgan v. V2X, Inc. (D. Colo. Mar. 30, 2026), where Judge Maritza Dominguez Braswell held that the identity of AI tools is not protected work product. There, the defendant raised alarms about whether confidential information, including trade secrets and personnel files, was being fed into mainstream platforms like ChatGPT or Gemini.

Judge Mitchell also argued that the ban can help discovery itself. She wrote: "If the court were to allow a party to upload all non-confidential documents and materials produced by another party to an open AI Tool, thereby making all such data amenable to public consumption, parties may err on the side of under-producing potentially responsive documents or seek to make extensive redactions of irrelevant or non-responsive information." She added that allowing closed AI tools "will facilitate discovery by incentivizing more fulsome document productions."

Everlaw, which sells ediscovery software, says the ruling highlights an evolving standard of care for data handling in the age of generative AI. For legal teams, the practical message in the post is that generative AI is welcome in discovery, but public, open-loop tools can carry risk, and closed, secure environments are often recommended when handling sensitive data.