NAIC Proposes Vendor Registry for AI Models Sold to Insurers

At the NAIC’s Spring 2026 National Meeting, regulators floated a proposed registry for vendors that provide AI models and datasets to insurers. The registry is not a licensure regime and would not relieve carriers of their existing vendor diligence and management obligations. Its purpose is to give state regulators more visibility into the third-party models and data sources flowing into insurance decisions.

The proposal matters because vendor oversight has become one of the most scrutinized parts of insurance AI governance. The NAIC has repeatedly stated that insurers remain fully responsible for third-party models, including those licensed from vendors or embedded in software. A registry would make it easier for regulators to identify which vendors are active in the market and to target examinations or information requests, but it does not change the underlying compliance burden.

For carriers, the practical implication is to prepare vendor files that could be submitted to a registry if the proposal moves forward. That includes model documentation, validation records, data source descriptions, and evidence of ongoing monitoring. Carriers that already have these materials organized for the NAIC AI Systems Evaluation Tool will be in better shape. Those that do not should treat the proposal as a deadline: a registry is easier to comply with when the documentation is already current.

The proposal also signals that vendor governance will remain a priority in underwriting and pricing examinations. Regulators want to understand not just what a carrier’s own AI program says, but who built the models it relies on and whether those vendors maintain appropriate governance. A registry would make that question easier to ask at scale.

For a checklist on AI vendor due diligence, see our guide to AI vendor risk assessment.