On December 1, 2023, the Innovation, Cybersecurity, and Technology (H) Committee (“H Committee”) met at the Fall 2023 US National Meeting of the National Association of Insurance Commissioners (“NAIC”). The meeting covered the following matters.
Adoption of the Model Bulletin on the Use of Algorithms, Predictive Models, and Artificial Intelligence Systems by Insurers
The H Committee adopted the Model Bulletin on the Use of Algorithms, Predictive Models, and Artificial Intelligence Systems by Insurers (the “Model Bulletin”), and it was later adopted by the Joint Meeting of Executive (EX) Committee and Plenary as well. During the H Committee meeting, regulators, trade organizations and consumer representatives discussed the latest draft of the Model Bulletin that had been exposed, including possible removal of the definition of “bias”—the most notable revision in the latest draft—and whether the remaining uses of the term should be clarified to mean “statistical bias” or potentially replaced with the term “unfair discrimination.” Ultimately, however, the H Committee adopted the Model Bulletin without further revisions.
The adopted Model Bulletin includes the following four sections:
- Section 1: Introduction, Background, and Legislative Authority: This section describes the existing insurance laws and regulations that govern insurers’ use of predictive models, artificial intelligence (“AI”) or machine learning, including the Unfair Trade Practices Act and the Unfair Claims Settlement Practices Act.
- Section 2: Definitions: This section defines certain terms used in the Model Bulletin. Notably, in the adopted version of the Model Bulletin, the term “bias” is not defined.
- Section 3 – Regulatory Guidance and Expectations: This section describes the expectation that all authorized insurers will develop, implement, and maintain a written risk-based program for the responsible use of AI systems that make or support decisions related to regulated insurance practices. This section also includes general guidelines for the development of such a program that covers general guidelines, governance, risk management and internal controls, and third-party AI systems and data. The Model Bulletin identifies robust governance, risk management controls and internal audit functions as powerful tools in mitigating the risk that decisions driven by AI systems will violate unfair trade practice laws and other applicable existing legal standards. It also encourages insurers to develop verification and testing methods to identify errors, bias and potential unfair discrimination in the use of predictive models and AI systems.
- Section 4 – Regulatory Oversight and Examination Considerations: This section includes a list of items of information and/or documentation that an insurance regulator may request when assessing an insurer’s use of AI systems.
Now that the NAIC has adopted the Model Bulletin, the next step is for individual state insurance departments to adopt and issue their own versions of the bulletin. How widely the Model Bulletin is adopted—and the extent of revisions made by the individual states—will be indicative of how much consensus exists among state insurance regulators on how insurers’ use of AI should be regulated. We will continue to monitor these developments.
Generative AI Presentation
The H Committee heard a presentation on generative AI from Victor Winter, a professor at the University of Nebraska–Omaha, who emphasized the exponential growth of technology and how quickly generative AI has developed and will continue to develop. Professor Winter also raised concerns about AI potentially being used in a manner that manipulates consumer decisions as well as the biases that may exist in generative AI systems.
To view additional updates from the US NAIC Fall 2023 National Meeting, visit our meeting highlights page.