Simon Glairy is a pivotal figure in insurance technology, bringing years of expertise in how artificial intelligence reshapes risk management and regulatory oversight. As the National Association of Insurance Commissioners (NAIC) prepares for its upcoming Spring National Meeting in San Diego, the conversation in the industry is shifting from theoretical AI risks to concrete, structured frameworks. This interview explores the development of the AI Systems Evaluation Tool, the rise of state-level modernization hubs, and the urgent data calls that will define the next five years of insurance oversight. We delve into how regulators are moving past outdated surveys to establish a direct, real-time understanding of predictive modeling and the evolving landscape of homeowners’ risk profiles.
The AI Systems Evaluation Tool is targeted for a 2027 launch with 14 states currently committed to its adoption. What specific technical benchmarks will this tool use to standardize insurer examinations, and how will it adapt to the different ways companies apply AI across various business lines?
The commitment of 14 states is a significant milestone for the 2027 launch because it demonstrates a collective desire for a unified, consistent framework. Rather than relying on sporadic feedback or self-reported narratives, the tool establishes a structured method that allows regulators to peer into the actual mechanics of company operations across different lines of business. We are looking at a system that moves away from the vague takeaways of the past and toward a technical benchmark that can be finalized for public comment by this coming September. This shift ensures that whether an insurer is applying AI in complex underwriting or high-speed claims processing, the examination process remains rigorous and predictable across different jurisdictions.
Existing surveys on AI adoption are often outdated by the time they are analyzed, creating a visibility gap for regulators. How will the shift toward direct, structured departmental inquiries improve your real-time understanding of predictive modeling, and what specific documentation should insurers begin organizing for these reviews?
The visibility gap has historically been a major hurdle because, by the time we analyze traditional survey data, the underlying technology has already evolved significantly. By shifting toward direct departmental inquiries, regulators can ask targeted, real-time questions that reveal exactly where and how AI is being deployed within a firm’s specific digital architecture. Insurers should proactively prepare by organizing documentation that details their predictive modeling processes and the internal governance surrounding these automated algorithms. This hands-on approach allows the NAIC to move beyond the limitations of the data collections seen earlier in 2024 and creates a dynamic feedback loop that keeps pace with rapid technological advancements.
Some state bureaus are establishing modernization hubs comprised of data analysts and cybersecurity experts to advise on emerging technology. How do these specialized units assist examiners who are less familiar with evaluating AI, and what are the primary challenges in building these multi-disciplinary teams within a regulatory framework?
Specialized units like the new insurance modernization hub in Virginia are essential because they provide a specialized “brain trust” to support traditional examiners who may not be data scientists. These units advise bodies like the State Corporation Commission on complex AI issues that might otherwise fall outside the expertise of generalist regulatory staff. The primary challenge lies in integrating these multi-disciplinary teams—which include strategic planners, cybersecurity experts, and policy advisors—into a rigid regulatory framework that wasn’t originally designed for high-speed tech iterations. However, these hubs are proving vital for evaluating predictive rate models, a task that has been identified as a brand-new capability for many seasoned insurance examiners.
A new homeowners insurance data call is expected to be conducted this year to update market insights. What specific shifts in homeowner risk profiles are driving this request, and how will the gathered data inform the goals of the upcoming national task force sessions in San Diego?
The data call expected this year is driven by the urgent need for fresh market insights that haven’t been comprehensively updated since the last major collection cycle. During the upcoming sessions in San Diego from March 22 to 25, the Homeowners Market Data Call Task Force will use this information to address the volatility in risk profiles that insurers are currently navigating. We expect the March 24 session to be particularly revealing as it sets the stage for how these new metrics will influence national policy and future rate-setting standards. By gathering this data now, regulators can better prepare for the structural shifts in the housing landscape before the 2027 AI evaluation tool becomes fully operational.
What is your forecast for AI regulation in the insurance industry?
My forecast is a transition toward a “living” regulatory environment where static rules are replaced by standardized evaluation tools that adapt as quickly as the code they monitor. Between now and the 2027 launch, we will see a surge in states joining the current 14 adopters as the benefits of a consistent, structured examination method become undeniable to policy leaders. We should expect the September finalization of these tools to trigger a massive wave of internal auditing within the industry as companies scramble to meet the new technical benchmarks. Ultimately, the success of this movement will depend on the continued collaboration seen at summits like the one held by the American Academy of Actuaries, ensuring that innovation and consumer protection move in lockstep.
