How do state insurance regulators oversee AI based pricing models

Building on Established Rate Regulation Practices

State insurance regulators oversee AI-based pricing models primarily by extending long-established insurance rate regulation practices — requiring insurers to file and justify their rating methodologies before use — to address newer, AI-specific concerns like algorithmic bias and proxy discrimination.

The Established Rate Filing and Review Process

Insurance rating methodologies have long been subject to a filing and review process with state insurance regulators, generally requiring insurers to demonstrate that proposed pricing factors are actuarially justified — meaning they have a legitimate, demonstrated connection to actual risk — before those factors can be used to set premiums for policyholders in that state.

Extending This Process to Address AI-Specific Concerns

As AI-based pricing models have become more prevalent, a growing number of states have extended and adapted this established review process to address concerns specific to these more complex, data-intensive models, including requiring insurers to provide documentation addressing how an AI model was developed, what data sources it uses, and evidence that the model doesn’t produce illegal discriminatory effects.

Requiring Specific Bias and Proxy Discrimination Testing

Some states have introduced specific requirements for insurers to test AI-based pricing models for potential bias and proxy discrimination — checking whether the model produces significantly different outcomes across protected groups, even when it doesn’t directly use a protected characteristic as an input — reflecting growing regulatory recognition of this documented risk.

Why Model Complexity Has Created New Regulatory Challenges

More sophisticated AI-based pricing models can be genuinely more difficult for regulators to fully evaluate compared to simpler, traditional actuarial models, since these models may incorporate many combined data factors and complex interactions that aren’t always straightforward to fully audit or explain, creating a genuine, acknowledged regulatory challenge as insurance pricing models continue to grow more sophisticated.

Why Specific Oversight Practices Vary by State

Given that insurance regulation operates primarily at the state level, the specific rigor and requirements of this oversight process vary considerably across different states, with some states having developed more detailed, AI-specific review requirements than others, reflecting the broader pattern of insurance AI regulation developing unevenly rather than through one uniform national standard.

Bottom Line

State insurance regulators oversee AI-based pricing models by extending established rate filing and review requirements to address AI-specific concerns, requiring insurers to demonstrate that pricing factors are actuarially justified and, increasingly, to provide specific documentation or testing addressing algorithmic bias and proxy discrimination — an oversight approach that continues to evolve as pricing models grow more sophisticated and complex to fully evaluate.

Go deeper

• How Do Insurers Use AI to Assess Risk and Set Premiums?
• Can AI Pricing Algorithms Charge Different Customers Different Prices?
• What is algorithmic transparency and why do regulators increasingly require it?