NAIC group nearly set to kick off pilot test to evaluate insurers’ use of AI

A National Association of Insurance Commissioners group moved closer Tuesday to launching a multistate pilot of a new artificial intelligence system evaluation tool this spring.

The Big Data and Artificial Intelligence Working Group addressed industry feedback during the call. Industry trade groups are critical of the effort.

The AI system evaluation tool is expected to enter a pilot phase in the coming weeks, with Colorado, Maryland, Louisiana, Virginia, Connecticut, Pennsylvania, Wisconsin, Florida, Rhode Island, Iowa and Vermont participating.

Insurance companies have steadily expanded their use of artificial intelligence over the past decade, moving from basic automation to more sophisticated applications throughout the insurance value chain. More recently, generative AI and advanced automation have reshaped how insurers engage with customers and manage internal operations.

Through it all, NAIC regulators have struggled to establish guardrails to preserve fairness, transparency and consumer protection. The working group proposed the AI Systems Evaluation Tool over the summer. Regulators are not required to use it, but it is another option when performing market conduct exams, regulators have said.

Regulators also said they plan to provide public updates during the spring as the pilot progresses, ahead of a summer national meeting. Officials acknowledged that some information may remain confidential but said they intend to share appropriate updates with stakeholders.

Industry feedback addressed

Representatives from trade associations, including the American Council of Life Insurers, the American Property Casualty Insurance Association and AHIP, provided comments on proposed edits and implementation details.

Karin Gyger, assistant vice president of insurance regulatory policy at ACLI, raised concerns about language in the tool, suggesting revisions to clarify that references to AI governance and risk assessment frameworks do not create new regulatory requirements.

"We will be revising that sentence based on your feedback," said Nathan Houdek, Wisconsin insurance commissioner and chair of the working group.

ACLI also requested the removal of the term “augment” from language describing AI models that “augment or automate” decision-making, arguing that the term could broaden the scope beyond direct consumer impact. Regulators said they plan to retain the broader language during the pilot phase to collect more comprehensive information, with the possibility of narrowing the scope after reviewing pilot feedback.

Dave Snyder, vice president, international policy and counsel, for the American Property Casualty Insurance Association, raised concerns about potentially duplicative information requests from multiple states. He also flagged questions about the use of terms such as “directly or indirectly,” ethical standards, language and definitions related to externally trained or generative AI models.

Regulators said they are seeking to balance broad information-gathering with clarity and efficiency.

"I think there are probably still places where direct, indirect is something that the regulators want to pose," explained Miguel Romero, director of property and casualty regulatory services at the NAIC. "Because that line can be blurry, and I think the regulators want to cast a wide net and then narrow that down as their inquiry continues."

Calls for coordination

Miranda Motter, senior vice president of state affairs and policy for AHIP, and Randi Chapman, managing director of state affairs at the Blue Cross Blue Shield Association, emphasized the importance of coordination among pilot states to avoid duplicative requests, particularly for insurers with complex holding company structures.

They also urged clarity around the pilot timeline, noting that insurers are currently focused on year-end financial statement filings. Regulators responded that the pilot is expected to begin after March 1 to accommodate those obligations, while acknowledging that individual states retain discretion over their participation.

Stakeholders also encouraged regular public reporting on the "lessons learned" during the pilot and urged consistency across participating states to ensure credible and comparable results.

Regulators said they are in the homestretch on edits to the tool, and invited stakeholders to submit specific redline proposals as soon as possible. A revised version reflecting agreed-upon changes will be circulated publicly before the pilot begins, Houdek said.

"We'll be creating a feedback loop, with the idea being that we're going to use this field testing process as a learning experience to both learn how we gather the information and the best process to do that, but then obviously, to understand what further refinements need to be made to the tool," Houdek told the working group.

© Entire contents copyright 2026 by InsuranceNewsNet.com Inc. All rights reserved. No part of this article may be reprinted without the expressed written consent from InsuranceNewsNet.com.

InsuranceNewsNet Senior Editor John Hilton has covered business and other beats in more than 20 years of daily journalism. John may be reached at [email protected]. Follow him on Twitter @INNJohnH.