The NIST AI Risk Management Framework (AI RMF 1.0) is a voluntary framework for managing risks across the AI lifecycle. It is the most practical starting point for most companies building AI governance.
The four functions
Govern
Culture, policies, roles and accountability for AI risk. This is cross-cutting — it informs the other three.
Map
Understand context: what each AI system does, who it affects, which data it uses and what could go wrong.
Measure
Analyze and track identified risks with quantitative and qualitative methods — testing, evaluation, monitoring.
Manage
Prioritize and act on risks: treat, transfer, accept or avoid, and respond to incidents.
The Generative AI Profile
NIST published a companion profile for generative AI that highlights risks such as confabulation, data privacy, information security and harmful content, with suggested actions mapped to the four functions.
Applying it in a SaaS company
- Build an inventory of AI systems and owners.
- Assess each system against the four functions.
- Prioritize gaps by business risk and customer demand.
- Map resulting controls into your SOC 2 program.
See our AI governance service for NIST AI RMF assessments.