Summary
Delaware's October 2024 interim policy is a durable legal-system baseline because it does not ban generative AI outright and does not treat it as casual office software either. The Supreme Court allowed judicial officers and court personnel to use approved GenAI tools, but only under a framework that keeps humans responsible for output accuracy, requires training and working knowledge, forbids delegating judicial decision-making, and blocks non-public information from non-approved tools. This is useful as an early, primary-source example of a court operationalizing AI with specific guardrails instead of abstract principles.
Why It Matters
This is a strong direct legal workflow record because it shows how a court turned general AI anxiety into concrete operating rules.
- it distinguishes approved from non-approved tools rather than talking about AI in the abstract
- it makes human accountability explicit even when AI is allowed
- it treats training and tool literacy as prerequisites for use
- it connects confidentiality and court-system infrastructure directly to approved-tool policy
What the Source Says
The Delaware Supreme Court's October 21, 2024 order adopted an Interim Policy on the Use of GenAI by Judicial Officers and Court Personnel as Appendix O-1 to the Judicial Branch Operating Procedures. The attached policy says GenAI is meant to provide assistance and is not a substitute for judicial, legal, or professional expertise. It defines "Approved GenAI" as tools approved by the Administrative Office for use on state technology resources and states that any use of output remains the responsibility of the authorized user. The policy requires working knowledge and training before use, prohibits delegating decision-making functions to GenAI, and bars users from inputting non-public information into non-approved GenAI.