The NIST AI Risk Management Framework, Explained in Plain Words
When a company starts using AI tools seriously, someone eventually asks whether there is a framework to follow. In the United States, a common answer is the NIST AI Risk Management Framework, usually shortened to the AI RMF. It sounds like a document for compliance specialists, and parts of it are. But its core ideas fit on one page and are useful for any team deciding whether and how to use an AI tool. The details below come from NIST's framework document and website.
1. What it is, and what it is not
The National Institute of Standards and Technology (NIST) released AI RMF 1.0 on January 26, 2023, as document NIST AI 100-1. NIST's website says it was developed through an open process that included a request for information, public comment drafts, and workshops.
Four words from the framework itself describe its character. It is intended to be voluntary, rights-preserving, non-sector-specific, and use-case agnostic. That means:
- No one is required to follow it by this document alone.
- It is meant for organizations of any size, in any industry.
- It applies to organizations that design, develop, deploy, or use AI systems, not only to those that build them.
As of October 2026, NIST's website says AI RMF 1.0 is being revised as part of the White House AI Action Plan, so expect updates. A companion Playbook offers suggested actions, and NIST has published additional profiles, including a Generative AI Profile in July 2024.
2. Part 1: what "trustworthy AI" means
The framework lists seven characteristics of trustworthy AI systems:
| Characteristic | Plain question to ask |
|---|---|
| Valid and reliable | Does it do the job correctly and consistently? |
| Safe | Could it put people in danger? |
| Secure and resilient | Can it withstand attacks and recover from problems? |
| Accountable and transparent | Is someone responsible, and can people see what it does? |
| Explainable and interpretable | Can users understand how and why it produced an output? |
| Privacy-enhanced | Does it protect personal information? |
| Fair, with harmful bias managed | Does it treat people equitably? |
Two details from the document help in practice. First, NIST describes valid and reliable as a necessary condition, the base the other characteristics rest on. A tool that gives wrong answers is not made trustworthy by being transparent. Second, NIST says trustworthiness involves tradeoffs: not every characteristic applies in every setting, and improving one can come at the cost of another. Its example is a possible tradeoff between interpretability and privacy.
3. Part 2: the four functions
The framework's "Core" organizes the work into four functions.
Govern. NIST describes this as cultivating a culture of risk management, with policies, roles and processes that connect AI use to the organization's values. It is cross-cutting: it applies at every stage and informs the other three.
Map. This function establishes the context: what the AI system is for, who uses it, who is affected, and what could go wrong. NIST notes that people working on one part of an AI system often lack full visibility into the others, which is why mapping matters.
Measure. This uses quantitative, qualitative, or mixed methods to analyze, benchmark, and monitor AI risks. NIST says AI systems should be tested before deployment and regularly while in operation.
Manage. This allocates resources to the risks that were mapped and measured, including plans to respond to, recover from, and communicate about incidents.
NIST notes that after putting governance in place, most users start with Map and continue to Measure or Manage, and that the process should be iterative. It also says the listed actions are not a checklist and not necessarily an ordered set of steps.
4. Using it on a small team
You do not need a risk department to borrow the structure. Here is one way a small team adopting an AI writing or meeting tool could apply the four functions:
- Govern: name one owner for AI tool decisions, write a one-page policy on what data may and may not go into AI tools, and decide who approves new tools.
- Map: for each tool, write down what it will be used for, who will rely on its output, what data it touches, and what a bad output could cause.
- Measure: test it on real examples before rollout. Track simple numbers, such as how often outputs need correcting, and check them again after a few months.
- Manage: decide what happens when something goes wrong, such as who is told, how errors are corrected, and when a tool is paused, and review the plan on a schedule.
The seven characteristics make a useful review checklist for step 2. If you cannot answer "Is it valid and reliable for this use?" with evidence, the Measure step is where to start.
5. Why it is worth knowing even if you never use it formally
The four function names and seven characteristics give a team shared words for AI risk questions. Knowing which characteristic a worry is really about, such as privacy or reliability, makes it easier to decide who should look at it and what evidence to ask for.
6. Key takeaways
- The NIST AI RMF 1.0, released January 26, 2023, is voluntary and meant for organizations of any size that build or use AI.
- It lists seven characteristics of trustworthy AI, with valid and reliable as the foundation.
- Its four functions are Govern, Map, Measure, and Manage; Govern applies throughout.
- NIST says its actions are not a checklist and the process should be iterative.
- A small team can apply it with a named owner, a short policy, real-world testing, and an incident plan.