NIST's AI Risk Management Framework is voluntary, and there is no fee or certificate attached to it. What it costs you depends on how many AI systems you have, how risky their uses are, how much governance you already run, and how deep you want to go. We give no prices. Here are the drivers to size your own effort.
What is the framework, in brief?
NIST describes the AI RMF as a framework for managing risks to individuals, organizations and society from AI, aimed at building trustworthiness into design, development, use and evaluation. It is intended for voluntary use. Version 1.0 came out on 26 January 2023, and a Generative AI Profile (NIST-AI-600-1) followed on 26 July 2024. NIST's page also says the framework is being revised as part of the White House AI Action Plan and mentions a critical infrastructure profile concept note dated 7 April 2026. Check the page for the current version before you start.
The four functions are Govern, Map, Measure and Manage. A companion playbook exists on NIST's Trustworthy and Responsible AI Resource Center.
Because it is voluntary, nobody issues a pass or fail. "Adopting" means you decide what to do, and you can show it. Some customers and contracts may ask you to align with it.
Driver 1: number and type of AI systems
Mapping an AI system means writing down its purpose, context and who it affects. Three internal tools take far less effort than thirty, and a hosted model used in a chat widget is simpler than a model you train and deploy yourself. Start by counting.
Driver 2: risk of the uses
A summarizer for internal notes and a system that influences decisions about people are not the same. Higher risk uses need deeper measurement and more review, so the effort rises quickly with risk.
Driver 3: governance already in place
The Govern function is about policies, roles and accountability. If you already have a risk committee, privacy review and security review, you can attach AI to them. If you have none, creating them is part of the cost and usually the slowest piece, because it needs decisions from leaders.
Driver 4: measurement and testing
Measure means choosing metrics and testing. The effort depends on what you test: accuracy, bias, resilience, security. Building evaluation sets and running them regularly takes engineering time and often needs domain experts.
Driver 5: third party models and data
Your suppliers are part of the risk. Reviewing a model provider, tracking updates to a hosted model you do not control and recording data sources all take time. The number of suppliers and how much they tell you set the effort. This is also where software supply chain practices matter: an inventory of libraries and models, and monitoring of advisories for them.
Driver 6: how deep and how often
A first pass that produces an inventory and a risk register takes weeks, not years, for a small company, but we cannot give you a timeline that holds for everyone. The more important question is how often you repeat it. Models, data and uses change, so a yearly refresh is a floor, not a ceiling.
Driver 7: tools and outside help
Some teams use spreadsheets, some use governance platforms, some hire advisers. Choose by the number of systems you track, not by feature lists.
Where Safeguard fits
Safeguard builds SBOMs and scans the software around your AI systems, which supports the inventory and the supplier monitoring parts. It supports Full AI, Zero AI or bring your own model for its own use, so you can choose how it handles code. It does not run your AI risk program or certify you against any framework.
What to do
- Count AI systems and sort them by risk.
- Name an owner per system and a group that reviews risk.
- Write a one page register: purpose, data, model source, owner, risk notes.
- Pick two or three measurements per higher risk system and run them on a schedule.
- Add model and data providers to supplier review.
- Re read the NIST page for revisions each quarter.
Quick check
Can you be certified against NIST AI RMF?
NIST describes the framework as voluntary and the page we read mentions no certification. Customers may still ask you to align with it.
What are the four functions?
Govern, Map, Measure and Manage, as named on NIST's AI RMF page.
Sources
- NIST: AI Risk Management Framework, fetched 18 October 2026
Last reviewed: 18 October 2026.
