Working tools · Open data
In insurance and financial services, the question is rarely can we build it — it's what will the regulator expect, and what will it cost us to satisfy them. I found the answers scattered across bulletins, law-firm alerts, and PDFs, so I built two tools to make them legible. Both are open — dataset, code, and methodology.
A live, sourced tracker of AI regulation in US insurance and financial services — state by state, framework by framework. Narrower and more specialized than the general trackers: every row links to its primary source and carries its own verification date. Built as an open dataset (CSV/JSON, CC-BY), not a law-firm content page.
Use it to answer: which states have adopted the NAIC AI Model Bulletin, which are in the evaluation-tool pilot, and which layer their own frameworks on top — plus the federal and EU picture.
Compliance is usually discussed as fear; this makes it arithmetic. An interactive view of what each regulatory regime actually costs to satisfy — so you can weigh the burden of a framework the same way you'd weigh any other product constraint.
Use it to answer: if we ship this model into these states, what does the compliance work actually look like — and where is the marginal cost of one more jurisdiction lowest.
A note on trust. Neither tool is legal advice. Every claim links to primary regulatory text — verify against the source before acting. That discipline is the point: in regulated AI, showing your sources is the product.