1. Home
  2. Blog
  3. Insurance

AI voice agents on insurance FNOL and claim status calls: what the agent must get right

Short answer

An AI voice agent on first notice of loss (FNOL) and claim status calls must verify the caller, state coverage only from the policy record, never promise an outcome, log a reported loss against the acknowledgment deadline, explain how a disputed decision gets reviewed, and disclose recording. State laws based on NAIC model law #900 and regulation #902 set the claims duties; privacy and wiretap law cover verification and recording.

On this page
  1. Which laws govern what an AI agent says on a claims call?
  2. How should the agent verify the caller before discussing a claim?
  3. What counts as misrepresenting a policy or promising an outcome?
  4. When does a phone call start the acknowledgment clock?
  5. What must the agent say when a caller disputes a decision?
  6. Do claims calls need all-party consent to record?
  7. What does the NAIC AI bulletin expect, and where has it been adopted?
  8. What does a graded claims call look like?
  9. Questions
  10. Sources

Your AI voice agent takes first notice of loss (FNOL) and claim status calls, and your team hears a sliver of what it tells policyholders and claimants. The failures that cost the most sound helpful: "you're covered" said without reading the policy, a claim "opened" after the tool call failed, a deadline started that nobody in claims can see. This guide is for the claims and contact-center leaders running these agents, their compliance officers and the engineers who build them. Most claims duties come from state unfair claims laws modeled on National Association of Insurance Commissioners (NAIC) models. Verification and recording rest on privacy and wiretap law, and the NAIC's AI bulletin adds expectations for the AI itself. States edit the models as they adopt them, and some go further, so check your own.

Which laws govern what an AI agent says on a claims call?

Your state's unfair claims settlement law, which in most states follows NAIC model law #900, the Unfair Claims Settlement Practices Act of 1990. Its section 4 lists fourteen unfair claims practices, from misrepresenting policy provisions to failing to explain a denial. Most day counts live in model regulation #902, for property and casualty claims; life, accident and health claims have their own, #903.

45

states with laws the NAIC treats as adopting model law #900

NAIC, Fall 2022

23

states that adopted regulation #902, plus 11 with related rules of their own

NAIC, Fall 2023

The NAIC counts California Insurance Code 790.03 and Florida Statutes 626.9541 among the #900 adoptions, and notes that states may not share its reading. It lists California's 10 CCR 2695 and Texas's claims rules as related to #902, not adoptions of it.

Two features of #900 matter for an AI agent. It creates no private right of action; the insurance department enforces it. And an act counts as an unfair claims practice only if it's committed flagrantly and in conscious disregard of the law, or "with such frequency to indicate a general business practice" (section 3). States differ on both. Florida lets a damaged person sue an insurer over 626.9541(1)(i) and waives the business-practice proof (Fla. Stat. 624.155(1)). Nevada makes an insurer liable to its insured for the unfair claims practices it lists (NRS 686A.310(2)).

The frequency test reads differently for software. Under the model, an adjuster who misstates a deductible once has made a mistake. Some states don't wait for a pattern. California counts a single knowing violation, and "knowingly" includes what the insurer should have known (10 CCR 2695.1(a), 2695.2(l)). Florida treats one material misrepresentation, made to settle a claim on less favorable terms than the policy provides, as a violation by itself (Fla. Stat. 626.9541(1)(i)2). Where a pattern is required, an AI agent makes it easier to reach. Its answers vary from call to call, but one bad instruction in its prompt can put the same misstatement on many of them. A pattern across many calls is what the frequency test looks for.

How should the agent verify the caller before discussing a claim?

Against the policy or claim record, by a tool call, before any claim detail, because claim files hold nonpublic personal information. For insurers, the Gramm-Leach-Bliley Act's privacy rules are enforced by state insurance regulators under state insurance law (15 U.S.C. 6805(a)(6)). In states that adopted the NAIC Insurance Data Security Model Law (#668), an insurer must also exercise due diligence in selecting any third-party service provider that can access that information, a voice vendor included. It must require that provider to protect the data (section 4F). Health coverage adds HIPAA's verification rule, covered in HIPAA and AI voice agents in patient access.

Model regulation #902 defines a claimant to include a designated legal representative and a designated immediate family member (section 3C), so your standard operating procedure (SOP) needs a rule per caller type:

One rule per caller, before any claim detail
  • Named insured: verified against the policy record by a tool call GLBA, state law
  • Designated family member or representative: the designation checked in the claim record, not taken on the caller's word NAIC #902
  • Third-party claimant: what they may learn about the insured's coverage varies by state, so a person answers that Your SOP
  • Attorney: routed to a person Your SOP

What counts as misrepresenting a policy or promising an outcome?

Section 4A of #900 prohibits knowingly misrepresenting to claimants and insureds relevant facts or policy provisions relating to coverages at issue.

A first-party claimant claims payment under the policy for a covered loss, usually the policyholder; a third-party claimant claims against someone the policy covers, such as the other driver in a crash. Model regulation #902 adds two duties on the call side. An insurer must fully disclose to first-party claimants all pertinent benefits, coverages or other policy provisions under which a claim is presented (section 5A). And it may not tell a third-party claimant their rights may be impaired if a form or release isn't completed within a set time, unless it's giving notice of a statute of limitations (section 7E).

A language model's failure here is filling a gap with a plausible answer. That's a hallucination, and how to catch AI voice agent hallucinations covers checking every such claim against the tool log. These rules catch most of it:

Rules that stop the agent filling gaps
  • Coverage, limits and deductibles stated only from what the policy lookup returned on this call, quoted where the wording matters NAIC #900
  • On a status call, the claim's status read only from the claim record fetched on this call Your SOP
  • No predicted outcome, amount or payment date: "an adjuster will review it" is a process, "you're covered" is a promise Your SOP
  • When a question goes past what the tools returned, the agent says so and hands off Your SOP
  • Denials left to a person with the file NAIC #900, #902

Model law #900 requires a prompt, reasonable and accurate explanation of a denial or compromise offer (section 4L), and regulation #902 a written denial citing the provision relied on (section 7A).

When does a phone call start the acknowledgment clock?

When it reasonably tells the insurer the facts of a claim and the policy accepts notice that way. First notice of loss (FNOL) is the policyholder's first report of a loss, often the first job an AI agent gets on a claims line. Under #902, a notification of claim is any notice, written or by other means acceptable under the policy, that reasonably apprises the insurer of the facts pertinent to a claim (section 3I). Notice to an insurer's agent is notice to the insurer (section 6A). Florida likewise treats a communication to an insurer's representative as one to the insurer (Fla. Stat. 627.70131).

The clock a phone report of a loss can start
  1. Day 0

    The loss is described to the agent, and notice has reached the insurer where the policy accepts it this way

  2. Promptly

    Claim forms, instructions and reasonable assistance go to the claimant (#902, section 6D)

  3. Within 7 days

    Florida property claims: reviewed and acknowledged, with a dated claim-file note if not in writing

  4. Within 15 days

    #902: acknowledged, and an acknowledgment not made in writing is noted, with its date, in the claim file

Source: NAIC model regulation #902, sections 6A and 6D; Fla. Stat. 627.70131 (7 calendar days)

So a loss described to the agent can be notice whether or not its claim tool worked. Where the policy makes written notice a condition, #902 lets the insurer deny for late written notice (section 5D), so the agent should say what must be sent in writing and by when. Check that the claim-creation call returned success, that a dated acknowledgment note was written and that forms went out. An agent that says "I've opened your claim" after a failed tool call has started a clock nobody in claims can see.

What must the agent say when a caller disputes a decision?

How the decision can be reviewed, and nothing that talks the caller out of it; the binding notices go in writing. For property and casualty claims, #902 creates no appeal right. Employer health and disability plans follow federal rules, and one of their clocks starts on the phone.

DeadlineWhat's owedSource
Once a claimant objects to a rejectionWritten notice that the state insurance department can review the matter, with its address and phone number#902, section 7H
At least 30 days before a limitation period expires (60 for third-party claimants)Written notice of the limit, before negotiating further with an unrepresented claimant whose rights it may affect#902, section 7D
5 days, or 24 hours for urgent careA group health plan explains the right pre-service procedure when a member's call names the patient, condition and treatment but doesn't follow it; orally is allowed29 CFR 2560.503-1(c)(1)
At least 180 daysThe time a group health plan must allow to appeal a denial2560.503-1(h)(3)(i)

A plan's written denial must describe its review procedures, time limits and the right to sue ((g)(1)(iv)), and most non-grandfathered health coverage adds external review, which denial notices must describe (45 CFR 147.136).

The SOP rule for a disputed decision
  1. Acknowledge

    Confirm the objection was heard, and never call a decision final while a review is open

  2. Explain review

    By the insurance department, or through the plan's appeal and external review

  3. Send the notice

    A tool triggers the written notice, which is the requirement; saying it aloud is good practice

On a health plan line, explaining the pre-service procedure before the call ends is an SOP choice that meets the 5-day rule on the spot.

Some states, including the three below, require everyone's consent, and a caller's location is rarely known.

WhereConsent to record
Federal law, 18 U.S.C. 2511(2)(d)One party's
California, Penal Code 632All parties to a confidential communication, or it's a crime; section 632.7 covers calls involving cell or cordless phones
Florida, Fla. Stat. 934.03(2)(d)All parties, given in advance
Washington, RCW 9.73.030All participants; a recording announcement counts as consent if the announcement is itself recorded

So the workable rule is a recording disclosure at the start of every call, on the recording. The AI vendor raises a second question. In February 2025 a federal judge let a California Invasion of Privacy Act claim proceed against Google over its Contact Center AI, which transcribed and analyzed calls to other companies' service lines (Ambriz v. Google). The court reasoned that a vendor capable of using call data for its own purposes can be an unauthorized third party. The order only let the claim proceed, but it's reason to read your voice vendor's data-use terms.

What does the NAIC AI bulletin expect, and where has it been adopted?

It expects insurers to govern the AI they use, vendors' systems included, under the laws that already apply. The NAIC adopted its Model Bulletin on the Use of Artificial Intelligence Systems by Insurers on December 4, 2023. It's framed as a reminder: decisions and actions that AI makes or supports must comply with existing law "regardless of the methods the Insurer used." That includes the unfair trade practices and unfair claims settlement practices acts.

What the NAIC AI bulletin expects of an insurer
  • A written AI systems program covering the whole insurance life cycle, claim administration included
  • Oversight of vendor-built AI too, with due diligence and, where appropriate, audit rights and regulator cooperation in the contract
  • Notice to consumers that AI systems are in use
  • Records an examiner may request, including validation, testing, auditing and model drift

As of August 31, 2026, the NAIC's adoption map lists the District of Columbia and 25 states as adopters. The states are Alaska, Arkansas, Connecticut, Delaware, Hawaii, Illinois, Iowa, Kentucky, Maryland, Massachusetts, Michigan, Mississippi, Nebraska, Nevada, New Hampshire, New Jersey, North Carolina, Oklahoma, Pennsylvania, Rhode Island, Vermont, Virginia, Washington, West Virginia and Wisconsin. California, Colorado, New York and Texas have their own insurance-specific AI regulation or guidance instead.

A voice agent that takes FNOL and claim status calls supports claim handling and is usually a vendor's system, so it's prudent to assume the bulletin reaches it. What an examiner would ask for looks like call QA: what the agent was supposed to do, how often it did, and whether a release changed that. Loops keeps that kind of record. It grades every call against rules the insurer's compliance lead approves and compares each rule's miss rate between agent releases. The claims rules it grades each cite the model section or statute they come from.

What does a graded claims call look like?

Like this, with each verdict tied to a timestamp or a tool result.

Example FNOL call, graded against a claims SOP(Illustrative)

Agent: Thanks for calling Cedar Mutual claims. I'm an automated assistant, and this call is recorded. What's your policy number?

1.1 met recording disclosed at 00:05, before any claim detail

1.2 met automated assistant disclosed at 00:04, per the SOP's AI-notice rule

Caller: AM-448120. A pipe burst in our kitchen last night.

Agent: I'm sorry to hear that. To confirm it's you, what's the ZIP code on the policy and your date of birth?

Caller: 33614, and April 9, 1971.

1.3 met verify_policyholder returned a match at 00:31

Agent: Thank you, Ms. Alvarez. I've opened claim 26-117045 for water damage at your home.

3.1 met create_claim returned success at 00:52; dated acknowledgment note written

Caller: Is this covered? I need to know before I call a plumber.

Agent: Sudden pipe bursts are covered under your policy, so you'll be fine. Go ahead and book the plumber.

2.1 missed coverage stated at 01:10; policy_lookup was never called

2.2 missed "you'll be fine" predicts the claim outcome

Agent: An adjuster will reach out within two business days.

2.3 needs review transcript can't show whether "two business days" matches the claim system's assignment rule

Agent: I'm also texting you a link with your claim forms and instructions.

3.3 met send_claim_packet returned success at 01:31

The agent got the procedure right, then answered the question a policyholder remembers. Coverage can turn on exclusions and facts no adjuster has seen yet, and the agent never read the policy. If the claim is denied, that recording is evidence of a coverage misstatement. If the same prompt produces it on many burst-pipe calls, that's the kind of pattern #900's frequency test describes, and in California one knowing misstatement is already enough.

This is general information, not legal advice; confirm with counsel how your states' versions of these models apply.

Write your claims SOP as numbered rules like these, each tied to its model section or state statute and to the tool call or quoted line that proves it. Then grade a month of real calls and look for rules missed at a rate, not once; our guide to auditing AI voice agent calls against your SOP covers the method. Loops grades every call this way and marks calls a transcript can't settle for review rather than guessing (how the grading works). A free 30-day audit grades last month's calls once your rules are approved.

Questions

Is a phone report of a loss valid notice if the policy asks for written notice?

Not necessarily. Under NAIC model regulation #902, a notification of claim is written notice or notice by other means acceptable under the terms of the policy (section 3I), and an insurer may deny a claim for late written notice when written notice is a written policy condition (section 5D). Log the call, tell the caller what the policy requires in writing and by when, and send the forms and instructions (section 6D).

Can policyholders sue under the Unfair Claims Settlement Practices Act?

Not under the NAIC model. It states that nothing in it creates or implies a private cause of action, and enforcement runs through the insurance commissioner's charges and cease-and-desist and penalty orders. State versions differ: Florida lets a damaged person sue an insurer over unfair claim settlement practices without proving a general business practice (Fla. Stat. 624.155), and Nevada makes an insurer liable to its insured for the practices its statute lists (NRS 686A.310(2)).

Does the NAIC AI bulletin require telling callers they are talking to AI?

The bulletin expects an insurer's AI program to include processes for notifying impacted consumers that AI systems are in use, with information suited to the stage of the insurance life cycle (guideline 1.9). It doesn't script a call opening. Saying so at the start of every call is the simplest way to show that notice happened, and it pairs naturally with the recording disclosure.

How long should an insurer keep AI claims call recordings and transcripts?

NAIC model regulation #902 counts all pertinent communications as claim documentation (section 3E), requires files detailed enough to reconstruct the insurer's activities (section 4B), and keeps a claim's key dates retrievable for the current year and the two before it (section 4A). States set longer terms: California requires claim files for the current year and the four preceding years (10 CCR 2695.3). Set voice-platform retention from your state's rule, not the model's.

Sources

  1. Unfair Claims Settlement Practices Act (Model Law #900), National Association of Insurance Commissioners
  2. Unfair Property/Casualty Claims Settlement Practices Model Regulation (#902), National Association of Insurance Commissioners
  3. Insurance Data Security Model Law (#668), National Association of Insurance Commissioners
  4. Model Bulletin: Use of Artificial Intelligence Systems by Insurers (adopted December 4, 2023), National Association of Insurance Commissioners
  5. Implementation of NAIC Model Bulletin: Use of Artificial Intelligence Systems by Insurers (status as of August 31, 2026), National Association of Insurance Commissioners
  6. Fair Claims Settlement Practices Regulations, 10 CCR 2695.1 to 2695.17, California Department of Insurance
  7. Florida Statutes 624.155, Civil remedy, The Florida Legislature
  8. Florida Statutes 627.70131, Insurer's duty to acknowledge communications regarding claims; investigation, The Florida Legislature
  9. 29 CFR 2560.503-1, Claims procedure (ERISA), eCFR, National Archives
  10. Ambriz v. Google, LLC, No. 23-cv-05437-RFL, Order Denying Motion to Dismiss (N.D. Cal. Feb. 10, 2025), U.S. District Court for the Northern District of California, via CourtListener