The NAIC Big Data and Artificial Intelligence (H) Working Group has issued important updates about its AI Systems Evaluation Tool and related pilot program. Following discussions on February 9 and 17, the working group announced that California has joined the pilot, increasing total participation to 12 states. The pilot began on March 2, 2026, and will run through September 2026. Below are key developments that insurance industry participants should be aware of.  

Background: The Rise of AI in the Insurance Industry

The rapid growth of big data and the adoption of artificial intelligence and machine learning (AI systems) are significantly transforming the insurance industry. These technologies may provide substantial benefits to both insurance companies and consumers by enabling the development of innovative products, improving customer services, simplifying and automating processes, and increasing efficiency and accuracy. However, without strong governance and effective controls, the use of AI systems may lead to negative consumer outcomes or threaten the financial stability of insurance companies. Insurers are responsible for managing the risks associated with developing and implementing AI systems and must show regulators that appropriate risk-based oversight mechanisms are in place and working effectively.

Recognizing these dynamics, the NAIC’s Innovation, Cybersecurity and Technology (H) Committee tasked the Big Data and AI Working Group with developing tools that enable regulators to continuously identify and evaluate risks related to AI systems. This initiative addresses both financial and consumer risks specifically arising from insurers’ use of AI systems. The AI Systems Evaluation Tool is designed to complement existing market conduct, financial analysis, and financial examination procedures for reviewing AI systems. These optional exhibits help regulators determine how extensively a company uses AI systems and whether additional analysis focused on financial and consumer risks is necessary.  

California Joins Pilot Program for AI Systems Evaluation Tool

The working group announced an expansion of the pilot program, which now includes 12 states: California, Colorado, Connecticut, Florida, Iowa, Louisiana, Maryland, Pennsylvania, Rhode Island, Vermont, Virginia, and Wisconsin. California joined most recently, following earlier announcements about Louisiana and Maryland.

The pilot will enable regulators to field-test the AI Systems Evaluation Tool. Participating states will use the tool for various tasks, including market conduct exams and reviews, financial analysis, and financial examinations. The initiative will involve insurance companies across different lines, such as property/casualty, life, and health insurance.  

Pilot Objectives and Implementation

The pilot is designed to accomplish several key objectives. Primarily, regulators want to determine whether the tool helps insurers clearly explain their AI governance systems, while also assisting regulators in better understanding how companies use AI systems and implement standard governance practices. The pilot will also support ongoing improvement and development of the tool itself, help create long-term recommendations for market conduct and financial risk assessment review processes, and identify what additional regulator training may be needed in the future.

During the pilot, participating states will concentrate on domestic insurers and implement a principle of proportionality. This means regulators will prioritize examining high-risk AI systems that could cause serious consumer or financial issues, while paying less attention to low-risk back-office systems.  

Updates in AI Systems Evaluation Tool Version 4.0

Version 4.0 includes several key updates:

Clarifying that the tool does not create new requirements for AI governance risk assessments

Simplifying prior direct/indirect impact references

Reinstating references to unfair trade practices for model testing inquiries

Adding a new data field in Exhibit D for “Reasonable Accommodations or Policy Modifications”

Remaining issues under review include materiality and risk assessment definitions, the inclusion of generalized linear models, and terminology for defining the scope of model inclusion.  

Pilot Framework and Timeline

Information requested through the tool will be protected under the confidentiality rules of the administering state. Participating states will receive training and coordinate through monthly calls to prevent duplicative requests. Regulators will provide public updates at each NAIC national meeting. From March to September 2026, pilot states will meet monthly to share progress. The tool will be updated based on pilot feedback in September–October 2026 and re-exposed for public review, with adoption expected at the NAIC fall meeting in November 2026.

As artificial intelligence reshapes the global economy, insurers face a fundamental shift in how risk is created, measured, and transferred. Speaking at an event hosted by MS Re during Miami Reinsurance Week, futurist Amy Webb outlined why the next decade will demand a new approach to risk.

The session, hosted by MS Re and attended by almost 200 insurance professionals, reflected growing industry focus on how emerging technologies could reshape risk over the coming decade.

The applications “may be a threat to some consulting businesses of insurance brokers though we view them as force multiplier rather than an existential threat,” he added.

Insurify’s app uses ChatGPT to compare auto insurance rates using details about the vehicle, the client’s credit history, driving record and other inputs. The company said the app launched on Feb. 3.

While the industry cut jobs in December, the pace of losses slowed from November, when insurance employment dropped by 9,200, according to the bureau.

Across the wider labor market, total nonfarm payroll employment increased by 50,000 positions in December and the unemployment rate remained relatively unchanged at 4.4%, the BLS said.

For 2025, payroll employment rose by 584,000, for an average monthly gain of 49,000 jobs, the bureau reported. In comparison, 2024 saw two million jobs created and an average monthly gain of 168,000.

The BLS reports total insurance payrolls each month on a seasonally adjusted basis alongside nonfarm payroll data. Industry segment figures - including employment across insurance carriers and noncarrier categories - are released separately on an unadjusted basis for the prior month.

Stephen Cooper, executive director and senior economist at the National Council on Compensation Insurance, said job gains in 2025 represented the slowest pace of annual job growth outside of a recession since 2003. Cooper said healthcare and social assistance added jobs during the year, while government jobs declined and manufacturing posted its third straight year of declines.

Cooper said 2025 was a "was a low-hire, low-fire environment where moderating but still solid wage growth offset sluggish employment growth to support overall payroll growth for workers' compensation."

He also said payroll growth was 4.3%, down from 4.6%, adding that payroll growth is a more important measure for workers’ compensation than unemployment data alone.

Separate industry research suggested many insurers still anticipate stable or higher staffing levels. In the third-quarter 2025 Insurance Labor Market Study released Aug. 19 by The Jacobson Group and Aon plc, 86% of insurance companies said they planned to increase or maintain staff levels in the next 12 months, including 53% expecting to add employees and 33% expecting to maintain staffing. Fourteen per cent (14%) reported plans to reduce headcount. The report projected industry employment could rise 1.03% over the next 12 months if carriers follow through on their plans.

The BLS said the January jobs report is scheduled to be released February 6.