Insurance agencies

Insurers are tightening up on AI. What agencies should do now.

More states have adopted the NAIC's AI bulletin, regulators are building a register of AI vendors, and carriers are filing AI exclusions. What it means for independent agencies and the businesses they insure.

Insurance is becoming the place where AI rules bite first. More states have adopted the NAIC's AI bulletin, regulators are working on a register of the vendors behind insurers' data and models, and carriers are filing exclusions that take generative AI out of standard liability cover. For independent agencies that means three jobs: get your own AI use written down, expect carriers to ask about it, and check your clients' renewals for new AI exclusions.

Agencies are using AI faster than they are governing it

The Big I's 2026 Agency Universe Study, published in September, counted about 37,000 independent agencies in the US, down from 39,000 in 2024. Nearly half (46%) now use AI, up from 15% two years earlier. Keeping pace with AI was the second most common challenge agencies named, at 43%, just behind finding good candidates at 45%.

Policies have not kept up. In the Big I Agents Council for Technology's survey in February, two-thirds of agencies planned to use more AI over the next year, but only 8% had it built into daily workflows and 55% had no written AI use policy.

Regulators are widening their view to vendors

The NAIC's model bulletin on insurers' use of AI systems sets out what regulators expect: a written AI program, governance, risk controls and oversight of third-party vendors. As of October 8, 2026, 29 jurisdictions had adopted it, and California, Colorado, New York and Texas have their own insurance-specific rules or guidance.

The next step is about vendors. An NAIC working group has been drafting a framework for third-party data and models used in property and casualty pricing and underwriting. The draft includes a shared multistate vendor register hosted by the NAIC, filing requirements, and annual attestations with notice of material changes. Regulators were considering bringing it to the NAIC's Fall National Meeting in November, along with an AI risk evaluation tool that 12 state insurance departments have been piloting in exams.

None of this applies to agencies directly. But the direction is clear: insurers will be asked to show what AI sits behind their decisions and who supplied it, and they will pass those questions down the chain.

Carriers are writing AI out of standard cover

In July, Insurance Journal reported that carriers are increasingly filing Verisk's optional generative AI exclusions with state regulators. CG 40 47 and CG 40 48 apply to commercial general liability and CG 35 08 to products and completed operations. The same report cited Gallagher's count of a 978% rise in AI-related lawsuits from 2021 to 2025.

Until now, many policies have covered AI risk silently, because nothing excluded it. That is changing at renewal, line by line. Gartner goes further, predicting that by 2030 insurers, not regulators, will drive AI governance through the standards they set for cover.

What agencies should do now

  1. Write your AI policy. One or two pages: which tools are approved, what client data can go into them, what a person must check before anything reaches a client or carrier, and what gets recorded.
  2. Keep a record of AI-assisted work. If an agent prepared a certificate, compared a policy or drafted an endorsement, you should be able to show what it saw, what it produced and who approved it.
  3. Read your own E&O renewal. Ask your E&O carrier how it treats AI-assisted servicing, and get the answer in writing.
  4. Check your clients' renewals for AI exclusions. Commercial clients using AI in their own businesses need to know if their general liability cover has changed. That conversation is good advice and good service.
  5. Be ready for carrier questions. A short written answer on how you use AI, how licensed decisions stay with people, and how you keep records will save time on every carrier review.

How we fit

Our servicing pack for independent agencies is built around those expectations. Licensed decisions always go to your staff, nothing goes to a client without review, and every step, by an agent or a person, is recorded in the Evidence File with a time and a name. If a carrier or regulator asks what happened, you can show them.

Common questions.

How many states have adopted the NAIC AI model bulletin?

As of October 8, 2026, the NAIC's adoption map lists 29 jurisdictions: 27 states plus the District of Columbia and Puerto Rico. California, Colorado, New York and Texas have their own insurance-specific AI rules or guidance instead.

Does the NAIC bulletin apply to insurance agencies?

It applies to insurers. Insurers are expected to oversee the third parties that support their processes, though, so carriers increasingly ask agencies and vendors how they use AI and what records they keep.

What are the Verisk generative AI exclusions?

They are optional ISO endorsements for commercial general liability and products and completed operations coverage, CG 40 47, CG 40 48 and CG 35 08, that exclude certain claims arising from generative AI. Insurance Journal reported in July 2026 that carriers are increasingly filing them with state regulators.

Do agencies need a written AI policy?

Yes. In the Big I's 2026 Agents Council for Technology survey, 55% of agencies said they had no written AI use policy. A short policy covering approved tools, client data, human review and record keeping is the first thing carriers, E&O insurers and clients will ask about.

Sources

Want this run for you.

Agentic MSP runs back-office work with governed AI agents and bills only for verified outcomes.