Insurance AI Under Fire for Alleged Racial Bias

What once looked like a niche civil rights lawsuit against a major insurer has emerged as part of a broader national fight over algorithmic decision-making in insurance. As companies rely more heavily on automated systems to detect fraud, process claims, and price risk, regulators and courts are beginning to ask whether those systems can unintentionally discriminate.

The growing number of complaints and lawsuits has prompted the insurance industry’s standard-setting organization to pilot a new tool to audit AI discrimination across the industry.

The lawsuit that first raised the question of AI discrimination in insurance

A federal judge in March declined a dismissal request in a 2022 federal class-action lawsuit against State Farm. The suit alleges that the insurer’s algorithmic claims-screening systems disproportionately subjected Black homeowners to greater scrutiny and delays.

The case is now in discovery, focused on State Farm’s automated claims tools.

When plaintiffs Jacqueline Huskey and Riian Wynn, who are Black, brought the claims, AI wasn’t yet the dominant public-policy story it became after late 2023. Since then, with a surge in litigation and scrutiny of automated claims-handling and underwriting algorithms, the case now looks more like an early bellwether than an isolated claim.

While State Farm denies the allegations, arguing that the lawsuit relies on unsupported allegations and that the plaintiffs failed to establish unlawful discrimination under federal housing law,  the judge declined to dismiss the suit, and the case is ongoing.

Questions spread beyond one case

This scrutiny isn’t limited to State Farm. Health insurers like Cigna, Humana, and UnitedHealthcare have also faced lawsuits over “batch-denial” algorithms that allegedly denied thousands of claims instantly, without meaningful human oversight.

In a May 2021 letter to the National Association of Insurance Commissioners (NAIC), the Center for Economic & Social Justice challenged an AI model some insurers use to evaluate claims. The model relies on data sources “known to be biased against communities of color,” the center stated.

That same year, an independent polling company, YouGov, surveyed roughly 800 white and Black homeowners with State Farm insurance.

Pollsters found “large and statistically significant racial disparities between Black and white homeowners.” Disparities spanned the time it took insurers to pay claims, the amount of paperwork needed to process the claims, and the number of interactions policyholders had with insurance representatives.

What’s next? NAIC taking action

Without acknowledging the lawsuits and complaints, the NAIC said in April it launched a pilot program for its AI Systems Evaluation Tool. The program is designed to test a framework that state regulators can use to assess how insurance companies govern and manage the risks of artificial intelligence.

The NAIC says the tool will help regulators understand insurers’ AI governance while allowing them to explain their systems clearly. Currently, 12 states participate in the pilot: California, Colorado, Connecticut, Florida, Iowa, Louisiana, Maryland, Pennsylvania, Rhode Island, Vermont, Virginia, and Wisconsin.

The tool’s main goal is risk identification. It helps regulators decide whether an insurer’s use of AI is “safe” enough to leave alone or requires an investigation to protect consumers from unfair treatment.

Last year, the NAIC surveyed health insurers and found that 92% use or plan to use AI, and nearly one-third don’t regularly test their models for bias or discrimination. This survey acted as a trigger for the current pilot.

This “adoption without testing” led regulators to worry about a systemic “black-box” risk in which unfair outcomes could occur without insurers even realizing it. Regulators intend the evaluation tool to address three main gaps in AI oversight:

• The “black-box” problem: Insurers can’t (or won’t) explain exactly how their AI systems make decisions. The tool creates a dialogue so insurers can clearly explain in terms that regulators understand.

• Regulatory fragmentation: The tool standardizes the inquiry, making compliance more predictable for insurers operating in multiple states.

• New risks (bias and drift): Standard audits aren’t designed to address AI risks such as “performance drift” (AI becoming less accurate over time) or “systemic bias.” The tool adds specific “health checks” to the audit process.

Following the pilot program, the NAIC plans to update the tool based on feedback in September and October 2026. The NAIC expects to officially adopt the finalized tool during the NAIC Fall National Meeting in November 2026.