Book a demo
Topics · AI-CONTROL-PLANE

Insurance AI Gateway NAIC Bulletin | Moring

The NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, adopted December 4, 2023, expects a written AI systems program with governance, risk management, and internal controls. An AI gateway is how carriers operationalize those expectations, producing per-decision records for underwriting, rating, claims, marketing, and fraud decisions.

On this page
  1. Short answer
  2. How it works
  3. FAQs
  4. Related reading
  5. Sources

Short answer

The NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, adopted December 4, 2023, expects a written AI systems program with governance, risk management, and internal controls. An AI gateway is how carriers operationalize those expectations, producing per-decision records for underwriting, rating, claims, marketing, and fraud decisions.

How it works

The NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers was adopted on December 4, 2023, as a template for state regulators. It creates no new substantive insurance law. Instead it signals how regulators intend to use existing unfair trade practices statutes and market conduct examination authority to review how insurers develop, acquire, and use AI systems that make or support consumer-impacting decisions. Its central expectation is a written AI systems program covering governance, risk management, and internal controls.

The bulletin applies across underwriting, rating, claims handling, marketing, and fraud detection, which makes it relevant to nearly every production AI use case an insurer operates. Controls are expected to be proportionate to the risk an AI system poses to consumers: a fraud-detection model used internally warrants different scrutiny than a pricing or underwriting model that directly affects policy terms. That proportionality principle maps naturally onto gateway route policy, where strictness can vary by workload.

Adoption has spread widely. As of October 2026, the NAIC's own map showed twenty-nine jurisdictions adopting the bulletin, with four states, California, Colorado, New York, and Texas, layering their own insurance-specific AI rules on top. Earlier in 2026 the count stood at twenty-five states. For a multistate carrier, the practical consequence is that bulletin-style expectations apply almost everywhere it writes business, and gateway-level evidence serves all of them at once.

In practice, bulletin compliance means an AI inventory, documented testing for unfair discrimination, vendor diligence, and internal controls that operate rather than merely exist. An enforcement AI gateway serves the inventory by making every model call visible, serves testing by recording which model version produced which decision, and serves internal controls by enforcing identity, redaction, and approval requirements at the request layer rather than in policy prose.

The gateway does not satisfy every bulletin expectation on its own. Validation work, fairness testing, and third-party diligence remain the carrier's obligation. But the bulletin's enforcement mechanism is market conduct examination, and examination asks for records of specific decisions. Gateway per-decision records, covering the data submitted, the policy applied, the model used, and the outcome, are precisely the artifacts an examining regulator requests when reviewing how an AI-supported underwriting or claims decision was reached.

Carriers should read the bulletin as an operating requirement, not a documentation exercise. Because it invokes existing unfair trade practices and rating and underwriting requirements, the test of compliance is whether AI-supported decisions would survive the same scrutiny as human ones. A gateway that binds identity, enforces policy, and records outcomes on every call is the mechanism that makes that demonstration routine instead of extraordinary.

Frequently asked questions

What is the NAIC Model Bulletin and when was it adopted?

The Model Bulletin on the Use of Artificial Intelligence Systems by Insurers was adopted on December 4, 2023, as a template for state insurance regulators. It establishes no new substantive law; it signals that existing unfair trade practices statutes and market conduct examination authority apply fully to AI-supported consumer-impacting decisions, and expects each carrier to maintain a written AI systems program with governance, risk management, and internal controls.

Which insurance functions does the bulletin cover?

Underwriting, rating, claims handling, marketing, and fraud detection, which spans nearly every production AI use case an insurer operates. Controls must be proportionate to consumer risk, so an internal fraud model warrants different scrutiny than a pricing model affecting policy terms. An AI gateway expresses that proportionality operationally by assigning different policy strictness, approval requirements, and logging depth to each route.

How many states have adopted the bulletin and what does that mean for carriers?

As of October 2026 the NAIC map showed twenty-nine jurisdictions adopting it, with California, Colorado, New York, and Texas maintaining their own insurance-specific AI rules on top; the count stood at twenty-five states earlier in the year. For a multistate carrier this means bulletin-style expectations apply almost everywhere it writes business, and a single gateway evidence layer can serve all of those jurisdictions at once.

How does an AI gateway help an insurer comply with the bulletin?

It operationalizes the program the bulletin expects: an inventory of model use through complete traffic visibility, records of which model version produced each decision, enforced identity and redaction controls at the request layer, and per-decision evidence covering data, policy version, timestamp, and outcome. Validation and fairness testing remain separate obligations, but the examination-ready records for specific decisions come from the gateway's normal operation.

Sources

  1. domain
  2. domain
  3. domain
  4. domain
  5. domain
  6. domain
  7. domain
  8. domain
  9. - artificialintelligenceact.eu
  10. - eur-lex.europa.eu