The NIST AI Risk Management Framework is voluntary US guidance, released in 2023, for managing AI risk across four functions: Govern, Map, Measure and Manage. It aims at trustworthy AI and has a Generative AI Profile companion for generative AI risks. It is a risk methodology rather than a certifiable standard, and it pairs naturally with ISO 42001, which provides the certifiable management system.
The NIST AI Risk Management Framework has become the common reference for thinking about AI risk in a structured way. It is not a law and not a certification, but it gives organisations a shared language and a method, which is why it turns up in almost every serious AI governance conversation.
What is the NIST AI Risk Management Framework?
The NIST AI Risk Management Framework, or AI RMF 1.0, is voluntary guidance from the US National Institute of Standards and Technology for managing risks across the AI lifecycle and promoting trustworthy AI. It was released in January 2023, and it has been widely adopted internationally as a structured approach to AI risk, well beyond the US.
Two things define it. First, it is voluntary and non-certifiable, guidance an organisation applies rather than a standard to be certified against, which distinguishes it from ISO 42001. Second, it is built around outcomes, trustworthy AI, rather than a fixed control list, which makes it adaptable to very different AI systems and organisations. The framework is organised into four core functions and mapped to a set of trustworthy AI characteristics, both covered below. Its role is to give an organisation a repeatable way to identify, assess and manage the risks of the AI it builds or uses, which is the foundation for any credible AI governance program.
What are the four functions of the NIST AI RMF?
The framework is organised around four functions, which together form a continuous cycle for managing AI risk.
| Function | What it covers |
|---|---|
| Govern | A cross-cutting culture and policy function: roles, accountability, policies and oversight for AI across the organisation |
| Map | Establishing context: scope, intended use, stakeholders, impacts and dependencies of an AI system |
| Measure | Analysing, benchmarking and monitoring risks using metrics, evaluations, testing and red-teaming |
| Manage | Prioritising and acting on risks: treatment, residual risk decisions, incident response, as a continuous loop |
Govern is the anchor, it runs across the other three and establishes the organisational culture and accountability that make risk management real rather than theoretical. Map, Measure and Manage then form the working cycle applied to each AI system: understand its context, measure its risks, and manage them, repeating as the system and its environment change. The cyclical design matters because AI risk is not static, models drift, uses expand, and the threat environment shifts, so the framework is built to be applied continuously rather than once.
What is trustworthy AI under the framework?
The NIST AI RMF frames its goal as trustworthy AI, defined through a set of characteristics an AI system should exhibit. These are the outcomes the four functions are meant to achieve, and they give the framework its direction.
The characteristics are that AI should be valid and reliable, safe, secure and resilient, accountable and transparent, explainable and interpretable, privacy enhanced, and fair with harmful bias managed. The breadth is deliberate: trustworthy AI is not just secure AI, it also covers fairness, transparency, privacy and safety, reflecting that AI risk spans more than cyber security. For an organisation, this means applying the framework is about more than protecting an AI system from attack; it is also about whether it behaves reliably, treats people fairly, and can be explained and held accountable. That wider lens is one of the framework’s strengths, and it is why it maps well onto broader AI governance rather than sitting only in the security domain.
What is the Generative AI Profile?
The Generative AI Profile, published as NIST AI 600-1 in July 2024, is a companion to the AI RMF that addresses the risks specific to or heightened by generative AI. It exists because generative AI, including large language models, introduced risks that the general framework did not fully anticipate.
The profile defines a set of generative AI risk categories, covering issues such as confabulation, the technical term for hallucination, data privacy, information security, harmful bias, information integrity, and the risks of relying on third party models and components. For each, it maps suggested actions back to the Govern, Map, Measure and Manage functions, so an organisation can apply the main framework specifically to its generative AI use. This makes the profile a practical bridge between the high level framework and the concrete risks teams face when they deploy chatbots, copilots and other generative AI, complementing the technical detail in the OWASP LLM Top 10. For an organisation adopting generative AI, the profile is a useful companion to the technical threat lists.
How does the NIST AI RMF relate to ISO 42001 and other frameworks?
The NIST AI RMF stands alongside several other AI frameworks, and understanding how they relate avoids treating them as competitors. The AI RMF is a voluntary risk methodology, setting out how to identify and manage AI risk. ISO 42001 is a certifiable management system standard, giving an organisation the system to govern AI, in the way ISO 27001 does for information security. The EU AI Act is binding law with penalties. These are layers, not alternatives.
In practice, many organisations use the NIST AI RMF for the risk methodology and ISO 42001 for the certifiable management system, with the two reinforcing each other, a comparison covered in ISO 42001 vs the NIST AI RMF. The AI RMF supplies a rigorous, respected way to think about and measure AI risk; ISO 42001 supplies the auditable governance structure and the certificate that customers and regulators increasingly ask for. For an Australian organisation, the AI RMF also aligns with the direction of national guidance, and it complements the ASD’s advice on securing AI systems. For an organisation building AI governance, Cybernion helps through ISO 42001 advisory and assessment.
Frequently asked questions
What is the NIST AI Risk Management Framework?
The NIST AI Risk Management Framework (AI RMF 1.0) is voluntary US guidance, released in January 2023, for managing risks across the AI lifecycle and promoting trustworthy AI. It is organised around four core functions, Govern, Map, Measure and Manage, and it maps to a set of trustworthy AI characteristics. It is guidance rather than a certifiable standard, and it is widely used as a structured way to think about and manage AI risk.
What are the four functions of the NIST AI RMF?
The four functions are Govern, Map, Measure and Manage. Govern is a cross-cutting function covering culture, policies, roles and accountability for AI. Map establishes the context, scope, intended use, stakeholders and impacts of an AI system. Measure analyses, benchmarks and monitors the risks using metrics, evaluations and testing. Manage prioritises and acts on the risks, including treatment, residual risk decisions and incident response, as a continuous loop.
What is trustworthy AI under the NIST framework?
The NIST AI RMF defines trustworthy AI through a set of characteristics that AI systems should exhibit, valid and reliable, safe, secure and resilient, accountable and transparent, explainable and interpretable, privacy enhanced, and fair with harmful bias managed. The framework’s functions are aimed at achieving these characteristics, so managing AI risk under the framework means working toward trustworthy AI rather than just avoiding technical failures.
What is the NIST Generative AI Profile?
The Generative AI Profile (NIST AI 600-1), published in July 2024, is a companion to the AI RMF that addresses risks unique to or heightened by generative AI. It defines a set of generative AI risk categories, such as confabulation or hallucination, data privacy, information security, harmful bias and information integrity, and maps suggested actions back to the Govern, Map, Measure and Manage functions. It helps organisations apply the framework specifically to generative AI.
How does the NIST AI RMF relate to ISO 42001?
They are complementary. The NIST AI RMF is a voluntary risk framework that provides the methodology for identifying and managing AI risk. ISO 42001 is a certifiable management system standard that provides the organisational system for governing AI. An organisation often uses the NIST AI RMF for the risk methodology and ISO 42001 for the certifiable management system, with the two reinforcing each other rather than competing.
Written by Gaurav Vikash, an ASD endorsed IRAP assessor and senior cyber security leader with 18 years of experience across Australia, the UK and Asia, including CISO and senior security leadership roles. He holds CISSP, CISA, CISM and CRISC and is an ISO 27001 Lead Auditor and ISO 42001 Lead Implementer, and speaks regularly at industry conferences.
Sources:
- AI Risk Management Framework, National Institute of Standards and Technology, 2026
- NIST AI 100-1, Artificial Intelligence Risk Management Framework (AI RMF 1.0), National Institute of Standards and Technology, 26 January 2023
- ISO/IEC 42001:2023, International Organization for Standardization, 2023
- Guidelines for secure AI system development, National Cyber Security Centre, 2023
Last updated: 30 August, 2026
