Skip to content
Judicial AI Standards Institute sealJudicial AI Standards Institute

Judicial AI Standard

Published - Issue 014

The Sandbox Is Not the System

Issue 014, Tennessee Edition · September 8, 2026

A sandbox is useful because it lowers the stakes of learning. It is not, by itself, a court operating model.

On August 26, the National Center for State Courts announced that the TRI/NCSC AI Policy Consortium had sunset its AI Sandbox. NCSC says the shift will let it focus more on educational and practical tools that increase trustworthy AI use, while it spends the next year developing common infrastructure to assess, research, and strengthen understanding of AI in courts.

That announcement should not be read as a Tennessee court rule, a procurement direction, or a signal that a court is ready to deploy a tool. It does, however, sharpen a practical question for Tennessee bench and bar: after people have explored a tool, what record, responsibility, and review process remain?

NCSC's readiness resource frames the answer as lifecycle work, not a one-time technology decision. It identifies the sequence from defining a workflow need and potential solution, through design and testing, implementation and evaluation, monitoring and updating, and post-project feedback.

Three records to carry forward

For a court or legal organization moving beyond informal exploration, three modest records can make that lifecycle visible:

  1. A use-and-boundary record. State the task, the users, the inputs that are permitted, the outputs that may be used, and the decision a human remains responsible for making.
  2. An accountable-owner record. Name the person or group responsible for reviewing performance, handling an error or complaint, and deciding whether the use should change, pause, or end.
  3. A feedback-and-change record. Preserve what the organization learned from testing or use: recurring errors, affected users, corrective changes, and the next review date.

Editorial guidance

Those are editorial recommendations, not a claim that NCSC or Tennessee requires a particular form. Their purpose is practical: an experiment can end with a demonstration; an operating practice needs a traceable way to explain what it does, who owns it, and what happens when the facts change.

For Tennessee judges and lawyers, the near-term discipline is straightforward. Do not treat “we tried it” as the end of governance. Ask what survives the trial: a defined purpose, a responsible reviewer, and a route for correction.

Direct official sources used

  1. National Center for State Courts, “NCSC's ongoing commitment to helping courts navigate AI”, Aug. 26, 2026.
  2. National Center for State Courts, “Understanding your court's AI readiness”, read Sept. 2, 2026.

The Bench & Bar AI Brief is an educational publication of the Judicial AI Standard project. It provides general information about artificial intelligence, legal ethics, court operations, and professional responsibility. It does not provide case-specific legal advice.