Book a demo
Topics · GOVERNANCE-AGENTS

Insurance Agent Identity NAIC Bulletin | Moring

The NAIC Model Bulletin adopted in December 2023 asks insurers to govern AI systems with written programs, named accountability, and examination-ready documentation. Agent identity controls give those expectations teeth: scoped per-agent credentials, named owners, and per-call evidence show an examiner that bulletin expectations for claims, underwriting, and service agents were enforced, not merely described.

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

Short answer

The NAIC Model Bulletin adopted in December 2023 asks insurers to govern AI systems with written programs, named accountability, and examination-ready documentation. Agent identity controls give those expectations teeth: scoped per-agent credentials, named owners, and per-call evidence show an examiner that bulletin expectations for claims, underwriting, and service agents were enforced, not merely described.

How it works

The Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, adopted by the NAIC in December 2023, states that decisions made or supported by AI must follow existing insurance law, including unfair trade practices and unfair discrimination rules. It expects every insurer to develop, implement, and maintain a written AI Systems Program for systems that make or support regulated decisions, and its examination section applies regardless of whether such a program exists. About 25 states had adopted the bulletin by April 2026, with several large states adding their own insurance-specific AI guidance.

Read against agentic deployments, the bulletin's expectations become identity requirements. Its accountability expectation, named owners for AI systems and outcomes, matches the identity governance principle that every agent carry a human owner, a business purpose, and a lifecycle state. Its risk-management expectation, controls proportionate to consumer impact, matches scoped credentials and tool-level policy: an agent handling claim outcomes needs tighter identity constraints than an agent drafting marketing copy. Its third-party expectation, insurer accountability for vendor AI, extends to the agent platforms, models, and tool providers an insurer's agents depend on.

The bulletin's examination section tells carriers what they will be asked: how they develop, deploy, and use AI systems, what outcomes those systems produced including adverse consumer outcomes, and what governance framework and internal controls applied. Compliance guidance for the bulletin maps these questions onto tool-call controls, recommending that carriers show, for a single tool call, the policy that applied and the decision made at that moment, confirm enforcement decisions are recorded as they occur, and retain approval records for actions requiring human sign-off. Identity is what makes those records coherent, because the actor on each record must be a scoppable, ownable agent rather than a shared credential.

Identity also resolves the bulletin's silent gap. The document was written for a world of models and decisions, not agents and tool calls, so controls enforced at each tool call produce the evidence an examiner would ask for while the mapping remains an insurer-documented interpretation. Agents acting in claims, underwriting, and customer communications are usually the highest-impact systems in the inventory, which makes their identity records the first place an examiner will look for the connection between the written program and observed behavior.

The bulletin's structure rewards insurers that treat identity as architecture rather than paperwork. It does not prescribe a technical standard; it expects carriers to explain how an AI system reached a decision, show that the system is monitored, and demonstrate accountability for its use. Per-agent identity, scoped grants, time-bound access, and recorded enforcement decisions satisfy those demonstrations across any adopting state's version. The EU AI Act and DORA align with the same traceability and resilience expectations for carriers with European exposure, so an identity architecture built for bulletin readiness generalizes.

Preparation follows the same order the bulletin implies. Inventory agents and their tools before anything else, because every other artifact depends on knowing what AI is in use. Write the identity rules into the AIS Program: who owns each agent, which permissions it holds, how permissions expire and are revoked. Then retain the per-call evidence that shows the rules ran. That is the difference between having a program and being able to prove one.

Frequently asked questions

What does the NAIC bulletin require that agent identity controls fulfill?

Four expectations: a written AI Systems Program, risk-based evaluation proportionate to consumer impact, defined accountability through named owners, and documentation retained for regulatory review. Scoped per-agent identities with named owners fulfill the accountability expectation directly, tool-level scoping makes risk-proportionate controls enforceable, and per-call records supply the documentation. The examination section applies regardless of whether a written program exists, so the evidence must exist too.

How should an insurer document its mapping of the bulletin to agents?

State the interpretation explicitly: the bulletin covers decisions or actions made or supported by AI systems, so its program, risk control, third-party, and documentation expectations extend to agents that act across claims, underwriting, and service workflows. Record which identity controls implement each expectation, such as named owners for agent permissions and retained approval records for consequential actions, and keep that mapping with the AIS Program.

Why is identity the weak point in bulletin readiness for agentic AI?

Because the bulletin's evidence expectations assume the carrier can say who acted. Agents provisioned with shared service accounts or inherited employee credentials break that assumption, leaving examination records that cannot tie an adverse outcome to a specific accountable actor. Per-agent identity with scoped, revocable credentials restores the chain from outcome to agent to owner, which is the chain the bulletin's documentation expectation implies.

Which bulletin provisions matter most for third-party agent platforms?

The bulletin makes insurers accountable for AI systems and models obtained from third parties, including monitoring for unfair discrimination. For agent platforms, that means diligence should cover how the vendor issues and rotates agent credentials, whether identities are distinct from administrative service accounts, and what enforcement evidence the platform produces. A governance gap in the vendor's identity model becomes the carrier's regulatory exposure during examination.

Sources

  1. content.naic.org
  2. content.naic.org
  3. www.vijil.ai
  4. www.moring.ai
  5. www.thedataexperts.us
  6. trussed.ai
  7. larsco.com
  8. www.thedataexperts.us
  9. - artificialintelligenceact.eu
  10. - eur-lex.europa.eu