Regulation · NIST AI RMF

Control AI risk with NIST AI RMF.

The NIST AI Risk Management Framework is a widely used, voluntary guide for trustworthy AI. We turn Govern, Map, Measure and Manage into controls that run on your AI, with evidence ready for any review.

  • Govern, Map, Measure and Manage applied to every AI system
  • Generative AI risks covered with the NIST AI 600-1 profile
  • One set of evidence that also supports ISO/IEC 42001 and the EU AI Act
MapMeasureManageGovernCROSS-CUTTING

The reality

You can't manage what you can't see.

Most teams know the framework exists. Few can show it working, because AI risk is spread across tools, teams and suppliers.

Visibility gaps

  • No list of where AI is used
  • Unknown third-party and embedded AI
  • No way to tell which uses are high-risk
  • Unclear owners and sign-off

Mapping needs a complete picture first.

Controls on paper

  • Policies nobody enforcesCommon
  • Manual, one-off checksCommon
  • No tests after go-liveRisk
  • Incidents handled ad hocRisk

A control that doesn't run isn't a control.

No proof

  • Which risks were assessed
  • What was tested and when
  • Who approved this system
  • What happened in the last incident

Evidence is what turns a framework into trust.

The framework

Four core functions.

The AI RMF Core organises AI risk management into four functions. Govern is cross-cutting; Map, Measure and Manage apply across the AI life cycle.

Govern

Culture, policies and accountability

Set the rules for how AI is used, who owns what, and how risk decisions are made. Govern runs across every other function.

  • AI policy and risk tolerance agreed
  • Roles, owners and sign-off defined
  • Third-party AI covered by policy
6 categories · GV
Map

Context and risks

Understand each AI system: its purpose, users, data, suppliers and the impact it could have on people and the business.

  • Purpose and users documented
  • Benefits, risks and impacts identified
  • Third-party models and tools recorded
5 categories · MP
Measure

Testing and metrics

Assess and track risks with quantitative and qualitative methods, before release and in production.

  • Accuracy, bias and robustness tested
  • Security and privacy risks evaluated
  • Metrics tracked over time
4 categories · MS
Manage

Treat and respond

Prioritise risks, act on them, and respond and recover when something goes wrong.

  • Risks prioritised and treated
  • Incidents tracked and resolved
  • Third-party risks monitored
4 categories · MG

One effort, three frameworks

Build once. Use the evidence everywhere.

The inventory, risk assessments, tests and logs you create for NIST AI RMF are the same building blocks other frameworks ask for.

Voluntary · US

NIST AI RMF

A flexible framework to identify and manage AI risk. Often requested by US clients and in procurement.

Certifiable · international

ISO/IEC 42001

A management system standard you can be certified against. Shares the same need for inventory, risk and controls.

ISO/IEC 42001 →
Law · EU

EU AI Act

Binding rules for AI in the EU. Your NIST evidence helps show risk management, testing and oversight.

EU AI Act readiness →

How we help

Each function, running on your AI.

We implement the framework on your governed AI layer, so the work is done continuously instead of once a year.

RequirementHow we cover itWhere it runs
GovernGV 1–6Policies set once and enforced on every AI request; owners, approvals and risk tolerance recorded.01Gateway04Run
MapMP 1–5Every AI system found, with purpose, users, data and third-party models documented in one registry.01Gateway04Run
MeasureMS 1–4Accuracy, bias, robustness and security tested before release and tracked in production.03ObserveSystem testingRed teaming
ManageMG 1–4Risks treated with guardrails, routing and approvals; incidents traced end to end.02Protect03Observe
Generative AI risksNIST AI 600-1Hallucination, data leakage, prompt injection and harmful content covered by tests and guardrails.02ProtectRed teaming

How it works

From framework to practice in five steps.

01

Baseline assessment

We review your AI use, policies and controls against the four functions and record your current profile.

Current profile
02

Set your target profile

Together we decide which outcomes matter most, based on your risks, clients and regulations.

Target profile
03

Implement controls

We put controls in place on your governed AI layer, from policies and approvals to testing and guardrails.

Controls running
04

Measure and report

Tests and metrics run continuously, with reports for management, clients and auditors.

Evidence and reports
05

Improve

Findings and incidents feed back into policies and controls, so the profile keeps moving forward.

Continuous improvement

The outcome

What you get.

Risks found early

High-risk uses flagged before deployment

Controls that run

Enforced automatically, not by checklist

Documentation ready

Versioned and traceable

Faster reviews

Less back and forth with clients and audits

FAQ

Frequently asked questions.

What is the NIST AI Risk Management Framework?

A voluntary framework published by the US National Institute of Standards and Technology in January 2023 to help organisations manage the risks of AI. Its core has four functions: Govern, Map, Measure and Manage.

Is it mandatory?

No. It is voluntary, but clients, partners and public buyers increasingly refer to it, and it is useful well beyond the US as a practical structure for AI risk.

Do we need to adopt every part of the framework?

No. The framework is designed to be adapted. You choose the outcomes that matter for your risks and context, and record them as a target profile.

What is the Generative AI Profile?

NIST AI 600-1, published in July 2024, is a companion profile for generative AI. It describes risks such as confabulation, data privacy and information security, with suggested actions.

Can we use it alongside ISO/IEC 42001 and the EU AI Act?

Yes. The frameworks overlap heavily. We build one set of controls and evidence and map it to each framework you need.

Contact us

Put NIST AI RMF into practice.

A 30-minute call with an engineer. We'll show where you stand across Govern, Map, Measure and Manage.

Cookie settings

Manage what we store on your device. Read the full Cookies Policy.

Strictly necessary

Required for the site and its forms to work. Always on.

Always on

Analytics

Google Analytics & Hotjar — traffic measurement and anonymised session insights.