United States · Voluntary framework
NIST AI RMF: map, measure, and manage AI bias with evidence
The NIST AI RMF 1.0 (January 2023) is a voluntary framework organised into four functions: Govern, Map, Measure, and Manage. It treats "fair, with harmful bias managed" as one of the characteristics of trustworthy AI. A Generative AI Profile (NIST AI 600-1) extends it to generative systems.
What the NIST AI RMF asks of AI systems
The obligations most relevant to bias, risk, and accountability, summarised from the NIST AI Risk Management Framework (AI RMF 1.0).
- GOVERNGovern
- Policies, roles, and accountability for AI risk across the organisation.
- MAPMap
- Context, intended use, and impacts of each AI system.
- MEASUREMeasure
- Evaluate trustworthiness characteristics, including fairness and bias (MEASURE 2.11), and document results.
- MANAGEManage
- Prioritise, respond to, and monitor identified risks.
Evidence FairMind helps you produce
FairMind links technical evidence to NIST AI RMF obligations for a human reviewer to accept. It is not legal advice and does not decide compliance.
- System records covering context and intended use (MAP)
- Evaluation runs and artifacts for fairness, safety, and robustness (MEASURE)
- Findings, remediation, and retests with owners (MANAGE)
- Reviewer decisions and policy links (GOVERN)
NIST AI RMF questions
Is the NIST AI RMF mandatory?
No. It is voluntary, but it is widely used as a reference by US agencies, enterprises, and procurement teams, and it maps well onto the EU AI Act and ISO/IEC 42001.
Where does bias testing fit in the NIST AI RMF?
Mainly in the Measure function. MEASURE 2.11 calls for fairness and bias to be evaluated and the results documented. The Manage function then covers responding to what you find.