AI Transparency in Insurance: What to Tell Clients

A client asks a simple question on a call: “Did a computer pick this carrier for me?” How you answer says more about your agency than any sales script. AI transparency in insurance is no longer an abstract policy debate. Regulators have written it into their expectations for insurers, and clients increasingly assume software is involved somewhere in the process. This article explains what AI transparency means for a licensed agent, where the regulatory expectations come from, and how to talk about AI tools with clients in plain, honest language.

What AI transparency means in insurance

Transparency is not the same thing as explaining how a model works. The National Institute of Standards and Technology (NIST) defines it in its AI Risk Management Framework as the extent to which information about an AI system and its outputs is available to the people interacting with it, whether or not they know they are interacting with one. NIST adds that meaningful transparency provides “appropriate levels of information” tailored to the role and knowledge of the person on the receiving end.

For an agent, that definition translates into three practical questions:

• Does the client know that software helped shape the recommendation?

• Could you describe, in one or two sentences, what the tool did and what you did?

• If the tool got something wrong, would you be able to trace it and fix it?

If the answer to all three is yes, you are already practicing transparency. If any of them is no, the rest of this article is for you.

Where the regulatory expectations come from

The National Association of Insurance Commissioners (NAIC) adopted its Model Bulletin on the Use of Artificial Intelligence by Insurance Companies in December 2023. According to the NAIC, the bulletin “establishes guidelines and expectations to ensure responsible use of AI by insurance companies” and “sets forth expectations as to how insurers will govern the use of AI.” The same NAIC page notes that state regulators “may require companies to explain how these tools are used in underwriting, pricing, marketing, or claims decisions.”

That last sentence matters for agents. A carrier that has to explain its AI use to a regulator will, sooner or later, expect its distribution partners to be able to explain theirs. The bulletin is addressed to insurers, not producers, but the direction of travel is clear. We covered the bulletin’s structure in more detail in what the NAIC’s AI rules mean for agents.

Adoption of AI in the life segment is still uneven. In the NAIC’s survey work summarized on that page, of the 161 life companies that responded, 58% reported using, planning to use, or planning to explore AI or machine learning models. The NAIC also reports that an AI Systems Evaluation Tool for regulators is being piloted by 12 states. The practical reading is that roughly four in ten responding life carriers were not yet on the AI path at the time of the survey, so the carrier on the other end of your case may or may not be using these tools in underwriting.

Transparency is different from explainability

Agents sometimes use the two words interchangeably. They are related but distinct, and the difference changes what you owe a client.

• Transparency is about disclosure: who used what, on which data, for what purpose.

• Explainability is about reasoning: why the tool produced this output instead of another.

NIST treats them as separate characteristics of trustworthy AI and makes a point that agents should remember: a transparent system “is not necessarily an accurate, privacy-enhanced, secure, or fair system.” In other words, telling a client you used AI does not make the recommendation right. It just makes the process visible. For the reasoning side, see our piece on explainable AI in insurance.

What to tell clients, and when

There is no single required disclosure script for producers today. What follows is a practical standard built from the NIST framing above and from how careful agencies already operate.

Say it early and plainly

A one-sentence disclosure during the fact-find avoids surprises later. Something like: “I use software that helps me compare carrier guidelines against your health history. I make the final recommendation, and the carrier makes the final underwriting decision.” That sentence covers the three questions from the first section and takes less than ten seconds.

Separate what the tool did from what you did

Clients care about who is accountable. Be specific. The tool narrowed a list, flagged a medication, or drafted a summary. You reviewed it, confirmed the details with the client, and chose the application to submit. Keeping those roles separate also protects you if a carrier outcome differs from the tool’s expectation.

Never let the tool speak for the carrier

A matching or quoting tool reflects published guidelines and the information the client gave you. It does not issue policies. Tell clients that any premium or class shown is an estimate and that the carrier’s underwriters decide the final offer. That wording is accurate and keeps the client’s expectations where they belong.

Be ready for “what did you tell it about me?”

Transparency about AI use leads naturally to questions about data. Have a short answer about what information the tool received and where it is stored. Our guide on client data and AI tools covers what agents can share with software vendors and what they should hold back.

A transparency checklist for the agency

Owners who want this to be consistent across a team can turn the standard above into a short checklist:

1. Every AI tool in use is listed somewhere a manager can see, with its purpose and the data it touches.

2. Agents have a one-sentence disclosure they actually use on calls, and it is in the call notes.

3. Recommendations carry a human sign-off before anything goes to a carrier.

4. There is a way to trace a tool output back to its inputs when a client or carrier questions it.

5. Tool outputs that turn out to be wrong are logged, not quietly corrected, so patterns show up.

Item four is the one most agencies skip. NIST notes that transparency “is often necessary for actionable redress related to AI system outputs that are incorrect or otherwise lead to negative impacts.” If you cannot reconstruct what the tool saw, you cannot fix the mistake, and you cannot explain it to anyone else. The same discipline that catches AI hallucinations in insurance work is what makes transparency real rather than a talking point.

How Peach Pilot approaches this

Peach Pilot’s Peach Quote is designed around the agent’s ability to explain each step. It compares a client’s stated health and lifestyle information against carrier guidelines and presents the results to the agent, who reviews them, discusses them with the client, and decides what to recommend and what to submit. We design for that visibility because an agent who can narrate the process is an agent a client trusts, and because the regulatory expectation is moving toward exactly that kind of traceability.

Frequently asked questions

Do I have to tell clients I use AI?

There is no single producer disclosure rule that applies in every state today. The NAIC Model Bulletin is addressed to insurers, and state adoption and additional requirements vary. The practical standard is to disclose anyway, briefly and early, because it costs nothing and builds trust. Check your state’s bulletins and your carriers’ producer guidance for anything more specific.

Is a quoting tool “AI” for disclosure purposes?

It depends on the tool. Rules-based filters and machine learning models are different technologies, but to a client the distinction rarely matters. If software materially shaped what you recommended, say so in plain terms.

What if the tool and the carrier disagree?

That is expected sometimes. Tools work from published guidelines and the information the client provided. Carriers apply full underwriting with records the tool never saw. Explain the difference, document it, and use the discrepancy to refine how you gather information next time.

Does transparency slow down the sale?

A one-sentence disclosure adds seconds. Discovering after submission that the client did not understand how the recommendation was made costs far more, in rework and in trust.

Conclusion

AI transparency in insurance comes down to a habit: tell clients that software helped, say what it did and what you did, and keep a record you could explain later. Regulators have made transparency a stated expectation for insurers, and the NIST framework gives everyone a shared vocabulary for it. Agents who adopt the habit now will find that the regulatory direction, the carrier relationship, and the client conversation all point the same way.

Peach Pilot supports licensed agents’ workflow. Carriers make final underwriting and issue decisions.