Judicial Council task force releases model generative AI use policy and outlines rule and standard development

The Judicial Council's Artificial Intelligence Task Force presented a model use policy for generative artificial intelligence (GenAI) on Feb. 21 and described companion proposals: a rule of court requiring courts that permit GenAI to adopt a conforming use policy and a standard of judicial administration addressing judges' use of GenAI in adjudicative roles.

Administrative Presiding Justice Mary Greenwood, chairing the presentation, said the model policy is intended to "balance responsible innovation that will benefit the courts and promote the administration of justice, but with safeguards that protect confidential information, ensure appropriate oversight, and maintain the public trust." The policy applies to court staff uses and judicial officers' non-adjudicative uses and would not require any court to permit GenAI.

Key provisions in the model policy described to the council include prohibiting submission of confidential or personally identifiable nonpublic information to public GenAI systems; requiring users to verify AI-generated material for accuracy and completeness; and a transparency requirement that GenAI-generated material that makes up a substantial portion of public-facing work include a disclosure identifying the system used. Greenwood and staff explained that the policy does not apply to internal systems accessible only from within a court's secure environment.

The task force said it surveyed courts in September 2024 and found many courts awaiting guidance; 42% of responding courts reported current GenAI use and 42% reported plans to use it. The task force is also developing a rule of court and a standard of judicial administration; both were described as planned to go out for public comment mid-March to mid-April and tentatively be considered by the council in July, with an anticipated effective date of Sept. 1, 2025, if approved.

Council members, including judges and court executives who have piloted internal closed-environment AI tools, praised the task force's work and raised issues for continued work — ethics guidance, training, bias, accuracy, procurement, and interaction with state technology guidance. The task force will continue development, including public comment and coordination with Department of Technology guidance and individual court pilots, and return with proposed rules and standards for council consideration.