NAIC Releases New Model Law for AI in Insurance Underwriting

The National Association of Insurance Commissioners (NAIC) has taken a significant step forward in regulating artificial intelligence in the insurance industry with the release of a new model law specifically addressing AI use in underwriting decisions.

Key Provisions of the Model Law

The model law, developed by the NAIC's Innovation, Cybersecurity, and Technology (H) Committee, establishes a framework for insurers using AI and machine learning algorithms in their underwriting processes. Key provisions include:

Transparency Requirements: Insurers must be able to explain how their AI systems make decisions in a manner that regulators can understand and evaluate. This doesn't require revealing proprietary algorithms but does mandate clear documentation of decision factors.

Bias Testing and Mitigation: Companies must regularly test their AI systems for unfair discrimination based on protected characteristics. The law requires annual bias audits and corrective action plans when disparate impacts are identified.

Human Oversight: While AI can assist in underwriting decisions, the model law requires meaningful human review of adverse decisions, particularly for coverage denials and significant premium increases.

Implications for State Insurance Departments

For state insurance commissioners and their examination teams, this model law presents both opportunities and challenges:

1. New Examination Protocols: Departments will need to develop expertise in evaluating AI systems, potentially requiring new staff training or external consultants.

2. Data Requirements: Regulators will need access to testing data and audit results, creating new data management considerations.

3. Synthetic Data Solutions: Many departments are turning to synthetic data to train examiners on AI evaluation without accessing actual consumer data—a privacy-preserving approach that Veridian specializes in providing.

What's Next

The model law is expected to be considered for adoption at the NAIC's next national meeting. States that adopt the law will have 18 months to implement the new requirements.

Insurance commissioners interested in preparing their departments for AI oversight should consider:

• Training staff on AI fundamentals and bias detection
• Developing examination checklists for AI systems
• Establishing relationships with technology experts
• Exploring synthetic data solutions for examiner training

The insurance industry's adoption of AI is accelerating, and regulators must keep pace to protect consumers while enabling innovation.