The National Association of Insurance Commissioners has extended a pilot program testing a unified artificial intelligence examination framework across 12 US states.
The organization's Big Data and Artificial Intelligence Working Group announced on August 13 that the trial of the AI Risk Evaluation Supplement will continue through September. State regulators use the document during insurance company examinations to structure inquiries regarding which artificial intelligence systems are deployed, how models are tested, and who holds operational responsibility.
The pilot includes regulatory agencies in California, Colorado, Connecticut, Florida, Iowa, Louisiana, Maryland, Pennsylvania, Rhode Island, Vermont, Virginia, and Wisconsin. Some state authorities have integrated the supplement into scheduled routine audits, while others are using it as a standalone questionnaire.
Under the current draft, insurance companies must describe their artificial intelligence applications across business lines, model governance structures, and validation procedures. Insurers must also detail high-risk models, data sources, monitoring processes, vendor controls, consumer impacts, and active safeguards. Regulators are required to tailor their requests to individual examinations rather than treating all questions as mandatory for every company.
Revision timeline and regulatory approval
Feedback from the ongoing trial will determine modifications for version 5.0, which working group leaders plan to release for a 30-day public comment period in September. Following a subsequent 14-day public comment period for version 6.0, the working group plans to consider version 7.0 for potential adoption at the NAIC fall national meeting.
Working group officials noted that timelines for individual state pilots vary, meaning the completion of all examinations by September 30 is not guaranteed. Officials emphasized that the August update remains a draft within the pilot process, meaning it does not constitute a mandatory nationwide requirement but shows how a unified state regulatory approach may take shape.
Impact on insurance technology teams
For machine learning teams developing systems in the insurance industry, the framework signals a growing need for technical documentation. State regulators may request up-to-date model inventories, records of model owners and administrative approvals, validation results, data origin history, monitoring logs, and documentation for third-party systems.
Insurance regulation in the United States operates primarily at the state level rather than through a single central federal agency. The National Association of Insurance Commissioners serves as the central standard-setting body governed by chief insurance regulators from all 50 states, the District of Columbia, and US territories.
