NAIC Renames AI Evaluation Tool to 'AI Risk Evaluation Supplement,' 12-State Pilot Runs Through September

The instrument pilot-state insurance regulators are using to examine insurer AI governance has a new name. The official meeting summary report from the NAIC’s Big Data and Artificial Intelligence (H) Working Group records the August 13, 2026 pilot update under the name AI Risk Evaluation Supplement. In the entry recording the group’s July 22 session, the summary pairs the new name with the old: “AI Risk Evaluation Supplement (formerly known as the AI Systems Evaluation Tool).” The summary does not say when the change happened. On the point that matters most to carriers, nothing moved: the pilot that began in March still involves 12 participating states and will continue through September.

The working group met on Thursday, August 13, from 1:00 to 2:00 p.m., during the NAIC 2026 Summer National Meeting in Columbus, Ohio.

The pilot update carried the session’s substance. Per the summary, pilot states are applying the supplement “in support of a mix of market conduct exams, financial exams, and financial analyses, as well as in the context of a more general regulatory inquiry.” Pilot state regulators are coordinating regulatory communication and determining which companies to include, and public updates will be provided throughout the pilot. The breadth of that description is the detail worth reading twice. A carrier in a pilot state cannot assume the supplement will surface only in a market conduct file; the summary places it across the financial tracks and general inquiry as well.

Formal action at the session was limited to housekeeping. The summary records the working group’s adoption of its July 22 minutes, which had in turn adopted the minutes of June 1. For the supplement itself, the summary records no adoption or exposure action at this session. That is a statement about what the official record shows, not a forecast of the working group’s next step.

The working group also received a presentation from AM Best on AI governance in insurance. In the rating agency’s account, governance should evolve as insurers move from predictive to generative and agentic AI, with governance proportional to the risk of each use case. The presentation emphasized validation, human oversight, monitoring, and documentation, pointed to controls such as authorization limits and rollback procedures, and said AI risks should be managed within existing enterprise risk management frameworks. Those are AM Best’s views as presented to the working group, not NAIC positions, and they are a useful signal of what a ratings analyst may soon ask about the same systems an examiner does.

For compliance teams, the rename is mostly administrative, but administrative is not the same as ignorable. Exam-preparation records, board materials, and vendor questionnaires that cite the AI Systems Evaluation Tool now point at a name the official record has retired. Updating the terminology is cheap today and confusing to leave until a regulator’s request letter arrives using the new one.

What the summary does not resolve is whether anything inside the instrument changed along with the name; it records a pilot update, not a new version. Background on the instrument and how it is built is in our standing analysis. The next fixed marker is the end of the pilot window in September, and the summary’s commitment to public updates through the pilot gives carriers something concrete to watch between now and then.