Insurance AI Regulation

    NYDFS AI Underwriting Guidance

    New York's Department of Financial Services expects insurers to prove their AI underwriting models are fair, explainable, and governed — not just documented. Regulators want evidence, not policy binders.

    Take the Assessment

    What NYDFS Requires

    Circular Letter No. 6 sets clear expectations for insurers using AI and external data in underwriting — with a focus on fairness, transparency, and accountability.

    Proxy Discrimination Testing

    Insurers must test whether AI models and external data sources produce unfairly discriminatory outcomes — even indirectly through proxy variables.

    Transparency & Explainability

    Underwriting decisions driven by AI must be explainable to regulators, consumers, and internal stakeholders in clear, non-technical terms.

    Board & Senior Oversight

    Governance of AI underwriting must include board-level accountability with documented oversight structures and escalation paths.

    Vendor Accountability

    Insurers are responsible for AI outcomes from third-party vendors and data providers — requiring due diligence and ongoing monitoring.

    The Reality of AI Underwriting Today

    Most insurers cannot explain how their AI models reach underwriting decisions — and regulators are no longer accepting that.

    Black-Box Models

    Complex ML models make underwriting decisions that even data science teams struggle to explain — leaving compliance teams without answers for regulators.

    Hidden Proxy Variables

    External data sources introduce proxy variables for protected classes — creating discrimination risk that traditional testing methods miss entirely.

    No Runtime Evidence

    Pre-deployment model validation doesn't capture what happens in production. Regulators want proof of ongoing governance — not one-time assessments.

    Expectation vs. Reality

    NYDFS expects continuous, demonstrable governance over AI underwriting. Most insurers are nowhere close.

    What Insurers Have

    • One-time model validation reports
    • Manual fairness testing at deployment
    • Vendor questionnaires instead of oversight
    • No audit trail of production decisions

    What NYDFS Expects

    • Continuous proxy discrimination monitoring
    • Explainability for every AI-driven decision
    • Board-level governance with documentation
    • Runtime audit evidence for examinations

    Real-Time Governance for AI Underwriting

    Trussed provides the governance infrastructure insurers need to prove compliance — not just claim it. Every AI underwriting decision is governed, logged, and audit-ready.

    Runtime Policy Enforcement

    Every AI underwriting interaction passes through Trussed's control plane — enforcing fairness, explainability, and data governance policies automatically.

    Examination-Ready Evidence

    Immutable audit trails capture every model decision, data source, and policy action — ready for NYDFS market conduct examinations.

    Continuous Fairness Monitoring

    Real-time dashboards track proxy discrimination risk, flag anomalies in underwriting outcomes, and surface issues before they become regulatory findings.

    Evaluate Your AI Underwriting Governance

    Understand where your organization stands on NYDFS AI underwriting compliance — and what gaps to close before your next examination.

    Take the Assessment