multi-agent-governance

NIST AI RMF Cross-Reference

This document maps the content in this repository to the corresponding NIST AI Risk Management Framework (AI RMF 1.0) functions, categories, and subcategories.

Full NIST AI RMF implementation guide: nist-ai-rmf-implementation-guide


NIST AI RMF Function Overview

┌─────────────┐     ┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│   GOVERN    │ ──► │     MAP     │ ──► │   MEASURE   │ ──► │   MANAGE    │
│             │     │             │     │             │     │             │
│ GV.1–GV.6  │     │ MP.1–MP.5  │     │ MS.1–MS.5  │     │ MG.1–MG.4  │
└─────────────┘     └─────────────┘     └─────────────┘     └─────────────┘

How This Repository Implements NIST AI RMF

GOVERN (GV) — Policies and Accountability

NIST Subcategory Implementation in This Repository
GV.1.1 — Organizational policies Governance framework documents serve as organizational policy templates
GV.1.3 — Prohibited uses defined Use case classification includes explicit prohibited patterns
GV.4.1 — Human oversight defined Human-in-the-loop requirements documented per pattern
GV.6.1 — AI systems monitored Monitoring requirements specified per design pattern

MAP (MP) — Risk Contextualization

NIST Subcategory Implementation in This Repository
MP.1.1 — Intended use documented Each pattern includes intended use scope and limitations
MP.2.1 — Scientific basis reviewed Patterns include evidence of effectiveness and known limitations
MP.3.1 — Risk identification Risk considerations documented per pattern/framework element
MP.4.1 — Impact assessment Stakeholder impact analysis included in high-risk patterns
MP.5.1 — Trustworthy AI characteristics Patterns mapped to NIST’s seven trustworthy AI characteristics

MEASURE (MS) — Risk Analysis

NIST Subcategory Implementation in This Repository
MS.1.1 — AI risk identification Risk categories enumerated with likelihood and impact
MS.2.3 — AI system monitoring Monitoring metrics and alerting thresholds defined
MS.3.1 — Evaluation techniques Evaluation approaches specified per framework element
MS.5.1 — Bias evaluation Fairness and bias considerations documented

MANAGE (MG) — Risk Response

NIST Subcategory Implementation in This Repository
MG.2.1 — Treatments defined Mitigation strategies specified for each identified risk
MG.4.1 — Rollback procedures Recovery and fallback procedures documented
MG.3.2 — Residual risk accepted Residual risk acknowledgment process defined

EU AI Act Alignment

For organizations subject to the EU AI Act, see the cross-reference mapping: nist-ai-rmf-implementation-guide/docs/eu-ai-act-mapping.md


The Seven Characteristics of Trustworthy AI (NIST)

Each component of this repository addresses one or more of these characteristics:

Characteristic Addressed By
Accountable Governance framework, role definitions, audit trails
Explainable Documentation requirements, decision logging patterns
Interpretable Output interpretation guidelines, confidence requirements
Privacy-Enhanced Data handling patterns, PII processing guidelines
Reliable Performance monitoring, regression testing requirements
Safe Safety evaluation checklists, failure mode analysis
Fair Bias evaluation requirements, subgroup testing

Maintained by Sima Bagheri · Not affiliated with NIST. For authoritative guidance, refer to airc.nist.gov