Skip to content
Judicial AI Standards Institute sealJudicial AI Standards Institute

The Bench & Bar AI Brief | Issue 006 | June 30, 2026

AI Governance Is Becoming Court Work

Rhode Island's competence amendment, NCSC and UNESCO judicial-training records, and recent appellate AI-citation cases show courts moving from warnings into practical governance.

Back to BriefOpen checklist

Tennessee Edition

Welcome to Issue 006 of The Bench & Bar AI Brief, Tennessee Edition.

Planned publication date: June 30, 2026.

For the past year, the easiest AI lesson for courts and lawyers was negative: do not file fake cases. That lesson still matters. But this week's source set shows the work is moving to a harder, more durable question: how should courts and legal institutions govern AI before the next mistake reaches a filing, order, training program, or public record?

The strongest signal is Rhode Island. The Rhode Island Supreme Court entered an order amending its lawyer competence comment to include the benefits and risks associated with existing and developing technology. The same order adopted interim GenAI guidelines for Rhode Island lawyers and for Rhode Island judicial officers. The order itself says those guidelines do not amend the professional conduct or judicial conduct articles and do not carry the force of law. That limit matters. Even so, the institution did not stop at a warning. It linked AI to competence, confidentiality, supervision, fees, public confidence, independence, integrity, impartiality, competence, and diligence.

That is the shape of the next phase. AI governance is becoming court work. It is not just a vendor issue or a tech-office issue. It reaches rules committees, judicial education, clerk training, bar guidance, procurement, records policy, confidentiality, and the ordinary professional duty to know what a tool can and cannot do.

The National Center for State Courts record points the same way. NCSC's AI Policy Consortium webinar, From theory to practice: A judge's hands-on guide to using AI, is described by NCSC as practical guidance for judges on using AI effectively, responsibly, and ethically in court operations and decision-making. UNESCO's June 26 public article reports training for more than 50 civic judges, mediators, and public defenders in Mexico City on ethical AI use, opportunities, risks, and safeguards. Singapore's Ministry of Law guide, verified from official government pages, uses the legal-sector principles of professional ethics, confidentiality, transparency, and human judgment. Different jurisdictions, same institutional pattern.

The enforcement record has not gone away. Recent verified court opinions from New York, Ohio, California, Iowa, Michigan, and other appellate courts show the same consequence line: fabricated citations, phantom cases, unsupported quotations, dismissal, sanctions, fee exposure, and disciplinary warnings. Those cases are not the whole story, but they explain why governance cannot wait until the filing is finished. A policy that only says verify citations is necessary. It is not sufficient.

Five signals to watch.

  1. 1Rhode Island tied technology risks to lawyer competence and adopted interim GenAI guidance for lawyers and judicial officers.
  2. 2NCSC framed AI use by judges as a practical court-operations and decision-making responsibility.
  3. 3UNESCO reported judicial-operator training on ethical AI use, safeguards, transparency, accountability, and human judgment.
  4. 4Singapore used professional ethics, confidentiality, transparency, and human judgment as legal-sector AI principles.
  5. 5Recent appellate AI-citation cases show why governance must happen before a filing, order, memo, or public record is finished.

A starter checklist.

For Tennessee courts, lawyers, and agencies, the question is practical. If a judge, clerk, lawyer, assistant, mediator, vendor, or contractor uses AI tomorrow, who decides what tools may be used, what data may be entered, what output must be checked, what records must be kept, what disclosures may be needed, and who is accountable when the work leaves the building?

  1. 1Treat AI competence as a training and supervision duty, not a one-time memo.
  2. 2Separate advisory guidance from binding rules, and label both clearly.
  3. 3Decide which court, client, agency, juvenile, health, financial, or sealed information may never enter unmanaged tools.
  4. 4Require independent verification of legal authorities, quotations, procedural claims, and factual assertions.
  5. 5Preserve prompts, outputs, model names, user decisions, and review notes when AI-assisted work may later be challenged.
  6. 6Build a human approval step for filings, orders, public notices, research memos, and public-facing explanations.
  7. 7Train judges, lawyers, staff, and vendors on the same plain rule: AI may help with work, but it does not hold the legal duty.

Bench & Bar takeaway.

The good rule is not "never use AI." The better rule is disciplined use. Courts and lawyers need a way to say yes to useful tools without surrendering confidentiality, candor, accuracy, records discipline, or public confidence. That is governance, and it now belongs on the bench-and-bar agenda.

  1. 1If a judge, clerk, lawyer, assistant, mediator, vendor, or contractor uses AI tomorrow, who decides what tools may be used?
  2. 2What data may be entered, what output must be checked, and what records must be kept?
  3. 3Does your office have an AI rule that a new lawyer, clerk, assistant, or vendor could actually follow tomorrow morning?

Distribution

Keep up between issues.

Subscribe to the weekly Brief by email, and follow Judicial AI Standard on X for daily AI and courts updates when verified information is available.

Subscribe to the BriefFollow on X
Return to The Bench & Bar AI Brief archive