Exhibit B
The NAIC AI Evaluation Tool's governance exhibit, answered as a narrative or a checklist, asking where an insurer's framework already covers each listed item.
Exhibit B carries the official title “AI Systems Governance Risk Assessment Framework,” and it comes in two forms an insurer can answer in: a narrative or a checklist. It sits downstream of Exhibit A, which a regulator uses first to decide whether the governance questions are worth asking at all.
The narrative form runs to five numbered questions. The first is the largest: the governance framework itself and who maintains it, board reporting and how often it happens, how the framework reaches the whole organization, accountability, and its integration with enterprise risk management and ORSA. The other four ask for the AI uses the company treats as material; testing and validation of models the company built and models it bought, with the frequency, scope, and method; the professional service providers it relies on, from actuarial and claim to MGA and audit; and anything else about how the framework is designed and assessed.
The checklist form asks a narrower set: whether a written AI Systems Program has been adopted, with the adoption date and how often it is reviewed, and whether the board or management was involved in adopting it. Then comes a single instruction to point at where the governance framework already addresses fourteen listed items. Those items: residual risk of unfair trade practices; compliance with state and federal law; adverse consumer outcomes; consumer data privacy; fitness for intended use; enterprise risk management; ORSA; the software development lifecycle; financial reporting risk; staff training; quantified AI risk levels; vendor procurement standards; complaint handling; and consumer disclosure.
The register matters more than the list. The preamble says the exhibit is “not intended to be interpreted as creating new requirements,” and the checklist tells you to reference where your framework already handles each item. Reading item 3k as an order to build a quantitative risk-scoring scheme, rather than as a place to say where yours lives, gets the exhibit backwards. All four exhibits are laid out in our NAIC AI Evaluation Tool guide.