Skip to main content

NIST AI RMF — U.S. Federal AI Risk Management Framework

Published 2026-04-19Updated 2026-06-308 min read

Overview​

NIST AI RMF (Risk Management Framework) is the AI risk management framework published by the U.S. National Institute of Standards and Technology (NIST) in 2023.

Key Features:

  • Voluntary Compliance — No legal enforcement
  • Federal Procurement Requirement: NIST AI RMF compliance mandatory for U.S. government contracts (EO 14110)
  • International Compatibility: Interoperable with ISO/IEC 42001

Version History:

  • v1.0 (Jan 2023): Initial release
  • v1.1 (Dec 2024): Added Generative AI section, enhanced transparency

4 Functions — GOVERN, MAP, MEASURE, MANAGE​

1. GOVERN​

Purpose: Establish AI system governance policies, culture, and accountability

Core Subcategories:

  • GOVERN-1.1: Establish AI risk management strategy
  • GOVERN-1.2: Clarify accountability (AI system owners)
  • GOVERN-1.3: Integrate legal, regulatory, and ethical considerations
  • GOVERN-1.4: Foster organization-wide AI risk culture

AIDLC Mapping: Governance Framework — 3-layer governance model

2. MAP​

Purpose: Understand AI system context, identify risks

Core Subcategories:

  • MAP-1.1: Understand business context (use cases, stakeholders)
  • MAP-1.2: Define AI system scope (inputs, outputs, dependencies)
  • MAP-2.1: Assess data quality
  • MAP-3.1: Identify risks (bias, privacy, security)
  • MAP-5.1: Impact assessment

AIDLC Mapping: Inception → Requirements Analysis, Reverse Engineering

3. MEASURE​

Purpose: Measure AI system performance, trustworthiness, fairness

Core Subcategories:

  • MEASURE-1.1: Define performance metrics (accuracy, F1, AUC)
  • MEASURE-2.1: Assess explainability
  • MEASURE-2.2: Bias testing (demographic parity, equalized odds)
  • MEASURE-2.3: Robustness testing (adversarial robustness)
  • MEASURE-3.1: Privacy impact assessment

AIDLC Mapping: Construction → Build & Test, Harness Engineering Quality Gates

4. MANAGE​

Purpose: AI risk response, monitoring, continuous improvement

Core Subcategories:

  • MANAGE-1.1: Execute risk mitigation strategies
  • MANAGE-2.1: Incident response planning
  • MANAGE-3.1: Continuous monitoring
  • MANAGE-4.1: Feedback loop (risk reassessment)

AIDLC Mapping: Operations → Post-market monitoring, incident response

NIST AI RMF 1.0 → 1.1 Major Changes​

Itemv1.0 (Jan 2023)v1.1 (Dec 2024)
Generative AIBrief mentionDedicated section added (Appendix B)
TransparencyMEASURE-2.1Enhanced (Model Card, Data Sheet examples)
Red Teaming-Added MEASURE-2.3 (adversarial testing)
Supply ChainGOVERN-1.5Expanded (open-source model risks)

U.S. Federal Procurement Requirements (EO 14110)​

Executive Order 14110 (Oct 30, 2023): "Safe, Secure, and Trustworthy AI"

Key Points:

  • Federal agencies must comply with NIST AI RMF when deploying AI
  • Models exceeding 10^26 FLOP must report to government
  • Federal procurement contracts must include AI risk management clauses

AIDLC Response: NIST AI RMF mapping mandatory for U.S. federal contract projects

AIDLC Integration Examples​

Inception Stage: GOVERN + MAP​

# .aidlc/compliance/nist-map.yaml
project: federal-contract-ai-tool
assessment_date: 2026-04-18

# GOVERN-1.1: AI Risk Management Strategy
governance:
  strategy: "Federal contract-compliant AI code generation tool"
  responsible_party: "AI Governance Team"

# MAP-1.1: Business Context
business_context:
  use_case: "Federal agency backend service code generation"
  stakeholders:
    - "Federal procurement officers"
    - "Development team"
    - "Security team"

# MAP-3.1: Risk Identification
identified_risks:
  - risk_id: RISK-001
    category: "Security"
    description: "Vulnerabilities in generated code"
    mitigation: "Automated SAST scanning"
  - risk_id: RISK-002
    category: "Privacy"
    description: "PII exposure"
    mitigation: "Guardrails filtering"

Construction Stage: MEASURE​

# .aidlc/harness/nist-measure-gates.yaml
quality_gates:
  # MEASURE-1.1: Performance Metrics
  - gate: performance_metrics
    enabled: true
    metrics:
      code_coverage: ">= 80%"
      duplication: "<= 3%"

  # MEASURE-2.2: Bias Testing
  - gate: bias_test
    enabled: true
    tests:
      - "demographic_parity_check"
      - "equalized_odds_check"

  # MEASURE-2.3: Robustness Testing
  - gate: adversarial_robustness
    enabled: true
    tools:
      - "bandit"  # SAST
      - "semgrep"

Operations Stage: MANAGE​

# .aidlc/monitoring/nist-manage.yaml
continuous_monitoring:
  # MANAGE-3.1: Continuous Monitoring
  metrics:
    - name: "error_rate"
      target: "< 1%"
      alert_threshold: 0.95
    - name: "bias_score"
      target: "< 0.05"
      alert_threshold: 0.04

  # MANAGE-4.1: Feedback Loop
  feedback_loop:
    frequency: "monthly"
    action: "Risk reassessment and mitigation strategy update"

References​

Official Documents:

Related Documentation: