What the NAIC's AI Rules Mean for Agents

Every few months a carrier announces something about artificial intelligence, and the question that reaches the field is the same one: does this change what I have to do on a case? The honest answer is that most of the regulatory activity so far has been aimed at carriers rather than producers. It still matters to you, because it shapes what carriers will ask for and how quickly a decision comes back.

Here is what has actually been adopted, what is still in progress, and what a producer should take from it.

Life insurers are the slow adopters, and the data says so

The National Association of Insurance Commissioners has surveyed insurers line by line on AI and machine learning. In the life insurance survey of 161 respondents conducted in December 2023, 58 percent reported current or planned use of AI or machine learning models.

That number looks lower once you see the other lines. Among 193 auto insurers surveyed in December 2022, 88 percent reported they use, plan to use, or plan to explore AI and machine learning models in their operations. Among 194 homeowners insurers in August 2023, the figure was 70 percent. Among 93 health insurers in May 2025, 92 percent said the same.

Life is the laggard, not the leader. If your carrier partners feel slower to automate than the rest of your financial life, that is not your imagination. It is the shape of the industry.

The Model Bulletin is about governance, not bans

In December 2023 the NAIC adopted a Model Bulletin on the Use of Artificial Intelligence by Insurance Companies. It is worth being precise about what a model bulletin is. It is not a law, and it does not prohibit particular tools. It sets an expectation that a carrier using AI in decisions affecting consumers can describe its governance: who owns the model, how it is tested, what happens when it produces an outcome nobody intended.

The practical translation is that carriers now have to be able to explain themselves. That tends to make them more conservative about automating a decline and more willing to automate the mechanical parts of a file.

A testing tool is coming

The NAIC has also been building an AI Systems Evaluation Tool for regulators. Per the NAIC’s own summary, it entered pilot testing with 12 states as of March 2026, with anticipated adoption at the 2026 Fall National Meeting.

Nothing in that is aimed at you. It does suggest the direction of travel: more documentation from carriers, and more questions about how an automated decision was reached.

What actually changes for the producer

Three things are worth internalizing.

• The recommendation is still yours. No governance framework transfers suitability responsibility from the licensed agent to a model. If a tool suggests a carrier and you present it, you own the presentation.

• Explainability cuts both ways. As carriers get better at explaining their decisions, you can ask better questions about a rating instead of accepting it as a black box.

• Clean inputs matter more, not less. Automated review is unforgiving about a file with gaps, which is the same failure mode we wrote about in why AI in insurance fails without clean data.

The honest read

Regulation here is early and mostly procedural. It is not going to change your Tuesday. What it signals is that the parts of the business most likely to be automated are the repeatable ones: intake, requirement tracking, matching a client profile against published guidelines. The parts involving judgment about a human being are exactly the parts nobody is rushing to hand over, which is the argument we made in what agentic AI in insurance actually automates and again in will AI replace life insurance agents.

Watch what your carriers do with the mechanical work. That is where the time comes back.

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