Human oversight in the creation of official verbatim court records required.
HF5155 would require human oversight whenever artificial intelligence-based tools are used to create official verbatim court records in Minnesota. The bill defines artificial intelligence-based tools broadly to include systems using machine learning, natural language processing, large language models, or similar technologies to generate, analyze, or transcribe court audio, video, or text for transcript purposes. It prohibits AI-generated recommendations, predictions, or outputs from being the sole basis for an official court record and requires review by a qualified court reporter.
Under the bill, human oversight must include real-time monitoring or post-process review, verification that the record accurately reflects the spoken words and context of the proceeding, and certification by the court reporter that the record complies with applicable law and standards. The Minnesota Supreme Court would be required to adopt or maintain policies governing this oversight before AI tools are used for official records. If there is a conflict between an AI-generated version and the human-verified record, the human-verified record controls. The bill takes effect January 1, 2027, and applies only to official verbatim court records created on or after that date.
The bill would create a new section in Minnesota Statutes chapter 484 and directly affect courts required to maintain official verbatim records, including the supreme court, district courts, and appellate courts. It would impose new procedural duties on courts and court reporters, including certification, documentation, and retention of records relating to AI-assisted transcription or recording. It also clarifies that unofficial recordings or transcripts are outside the bill’s requirements if clearly labeled and not used as part of the official record.
The available materials show no recorded committee debate or vote history, so there is no documented opposition or support to gauge from the legislative record provided. Based on the bill text alone, the measure appears to reflect a cautious, oversight-focused approach to AI use in the judiciary, emphasizing accuracy, accountability, and human control over official court records.
The main policy issue is whether AI can be used reliably in creating official court transcripts and, if so, how much human review is necessary. Potential points of contention include the cost and workload of requiring qualified court reporter oversight, whether the bill could slow adoption of transcription technology, and how broadly the definition of artificial intelligence-based tools should reach. Another possible issue is the extent of the Minnesota Supreme Court’s authority and responsibility to develop policies before AI tools are deployed in court record creation.