On December 4, 2023, the National Association of Insurance Commissioners (NAIC) adopted a model bulletin on the use of artificial intelligence (AI) in insurance. The impetus of the bulletin is the unique risks that AI poses to consumers, including “the potential for inaccuracy, unfair discrimination, data vulnerability, and lack of transparency and explainability.” The model bulletin requires insurers to adhere to several standards, including creating a risk-based framework to mitigate these risks and governance frameworks to ensure accountability and oversight of third-party vendors supplying AI systems.
As of October 1, 2024, 17 states have passed their own flavors of the regulation. While the bulk of these states adopted the NAIC model bulletin with negligible changes, some made more meaningful tweaks to the language. For example, Virginia’s guidelines implore insurers to “understand and eliminate” instead of “mitigate” the risk of adverse consumer outcomes presented by AI risks.
What this means for actuaries:
Actuaries are already quite familiar with the regulatory scrutiny around rating and use of generalized linear models in pricing. This bulletin extends the scrutiny beyond pricing models to every part of the insurance lifecycle that might impact the consumer: underwriting, claims handling, reserving and more! Wherever AI, machine learning or adjacent tools and techniques are used, the Department of Insurance will want to ensure proper steps are being taken to ensure interpretability, repeatability, reproducibility, traceability, robustness, regular tuning and monitoring of model drift.
As with most of our work, taking a state-specific lens will be necessary to understand the intricacies and strength of the language the states chose to adopt. Beyond the NAIC bulletin, four states struck off on their own and created their own legislation or regulation on insurers’ use of AI. While the enforcement of all these regulations is yet to be seen (though likely market conduct exam-triggered), working with your legal and compliance departments is your best bet. Keep an eye out, however, as additional states choose to either adopt the model bulletin or create rules of their own.
Sources:
- https://content.naic.org/sites/default/files/inline-files/2023-12-4%20Model%20Bulletin_Adopted_0.pdf.
- https://content.naic.org/sites/default/files/cmte-h-big-data-artificial-intelligence-wg-ai-model-bulletin.pdf.pdf.
Erin Lachen, FCAS, is a senior director II, data science at Liberty Mutual Insurance. She is a member of the AR Working Group and Writing Subgroup.
