On August 13, 2023, insurance regulators and the insurance industry met in Seattle, Washington at the NAIC Summer Meeting and continued to address the challenges of monitoring and regulating the insurance industry’s use of artificial intelligence. The NAIC Innovation, Cybersecurity, and Technology (H) Committee, as well as, the State of Colorado, are leading the efforts to produce regulatory frameworks for the insurance industry’s use of AI.

During the NAIC Summer Meeting, Colorado Insurance Commissioner Mike Conway spoke at a breakfast briefing sponsored by the American InsurTech Council. He discussed the status of Colorado SB 169 rulemaking (Proposed Algorithm and Predictive Model Governance Regulation (“CO Proposed AI Regulation”)) as well as the NAIC Exposure Draft of the Model Bulletin on the Use of Algorithms, Predictive Models, and Artificial Intelligence Systems by Insurers (“NAIC Draft AI Bulletin”). Commissioner Conway indicated that the current version of the CO Proposed AI Regulation will apply across all lines of insurance and will include a testing component. He also said that he was pleasantly surprised with the life insurance industry’s level of cooperation.

In addition, Commissioner Conway reported that he had just met with Google and Microsoft this past week as Colorado continues to understand how AI is being used in the insurance industry. When questioned about the NAIC Draft AI Bulletin, Commissioner Conway said that “it is going in a good direction.” States may diverge on the issue of testing, but Colorado wants insurers to test algorithms to prevent unfair discrimination. When questioned about the industry’s concern for protecting the confidentiality of AI data and information provided to Colorado, Commissioner Conway suggested that the new Colorado law includes confidentiality protections, but to the extent that an insurer continues to have concerns, he may consider accepting the information under the market conduct exam powers to provide additional confidentiality protection. Finally, while Commissioner Conway did not commit to an implementation date, he did indicate that given Colorado’s resources and other high priority concerns, for example, wild fires, it would be at least six months before the new regulation will be implemented.

Promptly following the American InsurTech Council Meeting, Commissioner Conway attended the NAIC Big Data and Artificial Intelligence (H) Working Group (“Big Data”) chaired by Rhode Island Insurance Commissioner Elizabeth Kelleher Dwyer. In addition to receiving a very comprehensive presentation from Deloitte on generative AI, Big Data also discussed the results of the recent Artificial Intelligence/Machine Learning survey conducted under the market conduct exam authority of 10 states (Connecticut, Illinois, Iowa, Louisiana, Nevada, North Dakota, Pennsylvania, Rhode Island, Vermont and Wisconsin). Of interest:

• With respect to homeowners insurers, 70% of the reporting insurers indicated that they “currently use, plan to use, or plan to explore using” AI as defined in the survey;
• With respect to private passenger auto insurers, 88% of the reporting insurers indicated the same;
• Below is the percentage of homeowners insurers using AI for certain functions as reported at the meeting:
  • 54% for claims;
  • 47% for underwriting and marketing;
  • 42% for fraud detection;
  • 35% for rating; and
  • 14% for loss prevention.

Later in the day, the NAIC Innovation, Cybersecurity, and Technology (H) Committee chaired by Maryland Insurance Commissioner Kathleen Birrane met in public session. Commissioner Birrane stated that 22 states had been working to produce the NAIC Draft AI Bulletin and stressed that it is not a model law or regulation, but merely is intended as interpretive guidance for the insurance industry. The (H) Committee received public comment from ten different industry and consumer groups. Generally, the insurance industry was supportive of the current NAIC Draft AI Bulletin, but acknowledged additional work is necessary to address concerns relating to, among other things, confidentiality, danger of unnecessary costs, definitions and glossary terms, ability of smaller insurers to negotiate with and monitor third party AI vendors, and testing. Consumer groups were critical of, among other things, the amount of time it is taking to produce a final product. Consumer groups also advocated for proxy discrimination to be defined. Consumer groups, like the State of Colorado, desire testing requirements. Finally, there was some discussion among regulators that the industry should include in their written comments a description of the industry’s concerns with respect to monitoring third party AI vendors for compliance and negotiating contract provisions with AI vendors to facilitate that monitoring.

The (H) Committee will continue accepting written comments on the NAIC Draft AI Bulletin through September 5, 2023.

Additionally, Colorado will conduct a virtual Permanent Rulemaking Hearing on Thursday, August 31, 2023 with respect to the CO Proposed AI Regulation.

Finally, on August 15, 2023, the Center for Insurance Policy and Research (“CIPR”) held a discussion on “Responsible AI/Big Data” featuring Maryland Insurance Commissioner Birrane and Vermont Insurance Commissioner Kevin Gaffney. There is an acknowledgment that there is a “chasm between what industry knows and what regulators know about AI and ML that needs to be closed. Regulators need training in these techniques to inform their regulation of artificial intelligence systems to protect consumers.” The panel discussed educational opportunities and the need to train and educate regulators on new technologies, including AI, data, analytics and modeling.  The Society of Actuaries and others described educational programs that are available in these areas. In responding to an interactive poll, respondents indicated that the biggest challenge to participating in such educational and training opportunities was finding the time to do so.

Overall, insurance regulators are continuing to develop a deeper understanding of how the insurance industry is using AI and its implications while grappling with how to best protect the insurance buying public without stifling the use of technological innovation.