Financial Services AI Risk Management Framework (FS AI RMF)
An operationalisation of the NIST AI Risk Management Framework tailored to financial services, version 1.0 dated 9 February 2026. Comprises an AI Adoption Stage Questionnaire, a Risk and Control Matrix, a User Guidebook and a Control Objective Reference Guide. The matrix contains 230 control objectives across four functions — Govern (81), Map (47), Measure (59) and Manage (43) — each paired with a named risk statement and mapped to maturity stages from Initial to Embedded.
Why it matters for autonomous finance
The most operationally specific document in agentic finance governance, and the least discussed. It is where abstractions become auditable line items: MG-2.4.3 requires technical mechanisms for system shutdown and disconnection, regularly tested; MP-3.5.3 requires documented criteria and thresholds for human intervention. Those are kill criteria and oversight, written by an industry body and ready to cite. We have cross-walked all 230 against MARQUE — see the mapping.
What it does not cover
Designed as a complement rather than a replacement for existing frameworks, and built for AI risk generally rather than for agents acting with capital. Of the 230 objectives, the overwhelming majority sit under a single principle — Accountable & Transparent — which tells you where the drafting energy went and where it did not. Coverage of delegated authority and agent identity is thin to absent.
No agent-specific authority model; no provenance or chain-of-title representation. Every entry in this register carries this section. A standard read past its scope is worse than no standard, because it produces confidence without coverage.