Pennsylvania's attorney general reached an agreement with GEICO on May 22 after an investigation found the insurer used an AI-enabled tool to select a new policyholder for additional underwriting review. The settlement signals that state regulators are watching how insurers apply automated decision-making, and it gives policyholders a template for pushing back when AI systems drive coverage decisions.
The investigation centered on GEICO's use of an AI tool during the underwriting process, a practice that is becoming common across the property and casualty industry. While the agreement's specific terms were not fully detailed in the initial announcement, the action establishes that insurers can face regulatory consequences when AI tools produce outcomes that state officials determine are problematic.
What the agreement means for underwriting practices
The Pennsylvania attorney general's office opened the investigation after receiving information about GEICO's AI-based selection process. The tool was used to flag certain new policyholders for further underwriting, effectively creating a second-tier review system that the state determined warranted scrutiny.
For insurers, the agreement is a reminder that AI deployment in underwriting is not exempt from existing insurance regulations and fair practices laws. Regulators are applying traditional standards to new technology, and they are willing to act when they see potential harm to policyholders.
How policyholders can respond to AI-driven decisions
Policyholders who believe an AI system played a role in an adverse coverage decision can take several steps. First, request a written explanation of the decision and ask specifically whether automated tools or algorithms were used in the review process. Insurers are increasingly required to disclose this information, and the question alone can trigger a more careful internal review.
Second, document everything. Keep records of application materials, correspondence with the insurer, and any notices about underwriting decisions. If a dispute escalates, a clear paper trail is essential. Third, file a complaint with your state's insurance department or attorney general's office if you believe the decision was unfair or discriminatory.
State regulators have shown they will investigate AI-related complaints, and each inquiry builds pressure on insurers to audit their systems for biased or arbitrary outcomes.
What to ask your insurer about AI use
Policyholders should ask direct questions about how AI is used in their specific case. Which tools were involved in the underwriting decision? What data did the system use? Was there human review of the AI's recommendation, and can you speak with the person who made the final call?
These questions are reasonable and increasingly common. Insurers that use AI tools for underwriting should be able to explain their processes without revealing proprietary information. If they cannot, that itself is a useful signal about the maturity of their governance practices. For professionals working in insurance operations, compliance, or risk management, understanding these regulatory expectations is becoming a core part of the job. Training in the legal and ethical dimensions of AI use can help insurance professionals identify problems before regulators do. See AI for Insurance for courses focused on these issues.
The legal framework around AI in insurance is still developing, and the GEICO agreement is one of the first concrete enforcement actions in this area. Insurance professionals who want to stay ahead of regulatory changes should track how state attorneys general and insurance commissioners interpret existing laws in the context of new technology. Resources covering the regulatory side are available through AI for Legal training programs.
Why this matters for insurance professionals
For underwriters, claims adjusters, and compliance officers, the GEICO agreement is a practical warning: AI tools do not immunize an insurer from liability, and regulators will hold companies responsible for outcomes, not just intentions. Anyone involved in building, purchasing, or overseeing AI systems for underwriting should verify that those systems include human oversight, documented appeal processes, and clear explanations for adverse decisions.
The cost of getting this wrong is not hypothetical. Regulatory actions like this one create reputational damage, legal expenses, and potential liability for policyholders. Insurance professionals who can audit their own AI systems for fairness and transparency will be the ones who protect their companies from the next investigation.
Your membership also unlocks:
