NIST AI Risk Management Framework
A voluntary framework organized around four functions — Govern, Map, Measure, Manage — for identifying and managing risk across an AI system's lifecycle. Published January 2023, with a generative AI profile added subsequently.
Why it matters for autonomous finance
The reference document US financial institutions will be measured against, whether or not it is ever mandated, because it is what examiners already know. Its real contribution is the Govern function: it puts organizational accountability ahead of technical controls, which is the correct ordering and the one most firms invert.
What it does not cover
It is a process framework, not a conformance standard. There is no certification, no pass mark, and no defined threshold — two firms can both claim alignment while operating at completely different risk levels. It also predates autonomous agents acting with capital, and reads throughout as though a human decides and the model advises.
No treatment of delegated authority, agent identity, or systemic correlation. 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.
Read it
Other standards in the register
Source and provenance
- Issuing body
- US National Institute of Standards and Technology
- Canonical source
- https://www.nist.gov/itl/ai-risk-management-framework
- Version
- AI RMF 1.0
- Dates
- published January 2023
- Last verified
- Next review due
- Recorded claims
- nist-ai-rmf (2026-08-02)
- Content held
- This corpus holds no copy of the source's text; the record is a citation, and no content digest can be given.
- Third-party material
- The framework, its text, identifiers and version belong to its issuing body (US National Institute of Standards and Technology). This corpus cites it, reproduces none of its text, and grants no rights in it.
- This corpus's work
- The summary, why-it-matters, caveat, gap statement and MARQUE mapping on this page are this corpus's own editorial work, under the terms at https://qistrust.com/license/.
Dates are when each recorded claim was last checked against its primary source (the verification register), not when this build ran; last verified is the oldest of them. The same record, field for field, is in api/provenance.json (qis-provenance-1) and in this page's structured data.