Summary US state insurance regulators are pressing ahead with a 2026 pilot of an artificial intelligence systems evaluation tool involving ten insurers, over trade-group objections that the pilot is voluntary for regulators while compulsory for participating companies, with no defined duration.
Source: InsuranceNewsNet · 12 December 2025 · read the original article
Why the pilot design is contested The industry objection is procedural rather than philosophical. A pilot with no defined end date and asymmetric participation obligations creates open-ended examination exposure: findings may be used, scope may expand, and participants have limited ability to withdraw. Those concerns are familiar from earlier market conduct pilots and are usually resolved by publishing a scope document and a sunset date.
The substantive question is more interesting. Whatever the tool asks will become the de facto template for standardised state AI examinations of insurers, because regulators reuse instruments that work.
What the tool will likely standardise Based on the NAIC model bulletin lineage, an evaluation instrument is likely to seek an inventory of AI systems by function, governance documentation and accountable owners, evidence of testing for unfair discrimination, third-party model oversight, and consumer-facing disclosure and appeal routes.
Insurers not in the pilot should prepare against that structure now. Assembling the evidence retrospectively during an examination is materially more expensive than maintaining it.
The preemption fight A December 2025 executive order pursuing a single national artificial intelligence regulatory framework put the state model directly in question, with Iowa's commissioner arguing it would undercut 150 years of state regulatory authority. The outcome determines who runs AI examinations of insurers, not whether they happen.
For planning purposes the prudent assumption is dual-track: build to state expectations, which are concrete and imminent, while keeping the programme documentation portable enough to satisfy a federal framework if one arrives.
Methodology and limitations This summarises trade reporting as at the date shown and links to the original. Pilot scope, participation and the status of federal action have continued to develop, and the described tool content is inference from published NAIC work rather than a released instrument.
Related reading See [Insurance Risk](/expertise/insurance-risk), [Regulatory Compliance](/expertise/regulatory-compliance), [Governance, Risk and Compliance](/expertise/grc) and the overview in [AI governance in insurance and banking](/insights/ai-governance-insurance-banking-2026).
Frequently asked questions
Why the pilot design is contested?
The industry objection is procedural rather than philosophical. A pilot with no defined end date and asymmetric participation obligations creates open-ended examination exposure: findings may be used, scope may expand, and participants have limited ability to withdraw. Those concerns are familiar from earlier market conduct pilots and are usually resolved by publishing a scope document and a sunset date.
What the tool will likely standardise?
Based on the NAIC model bulletin lineage, an evaluation instrument is likely to seek an inventory of AI systems by function, governance documentation and accountable owners, evidence of testing for unfair discrimination, third-party model oversight, and consumer-facing disclosure and appeal routes.
What should risk leaders know about the preemption fight?
A December 2025 executive order pursuing a single national artificial intelligence regulatory framework put the state model directly in question, with Iowa's commissioner arguing it would undercut 150 years of state regulatory authority. The outcome determines who runs AI examinations of insurers, not whether they happen.
What should risk leaders know about methodology and limitations?
This summarises trade reporting as at the date shown and links to the original. Pilot scope, participation and the status of federal action have continued to develop, and the described tool content is inference from published NAIC work rather than a released instrument.
What should risk leaders know about related reading?
See [Insurance Risk](/expertise/insurance-risk), [Regulatory Compliance](/expertise/regulatory-compliance), [Governance, Risk and Compliance](/expertise/grc) and the overview in [AI governance in insurance and banking](/insights/ai-governance-insurance-banking-2026).