NIST AI Risk Management Framework (AI RMF 1.0)

What it is

The Artificial Intelligence Risk Management Framework, published as NIST AI 100-1, was released by the US National Institute of Standards and Technology on 26 January 2023. It was developed through an open, consensus-driven process with industry, academia and civil society, and it is free to use.

The framework has two parts. Part 1 frames AI risk, describes the audiences involved across the AI lifecycle and sets out the characteristics of trustworthy AI. Part 2 is the Core, four functions broken into categories and subcategories of outcomes. NIST also publishes the AI RMF Playbook, which gives suggested actions, references and documentation for each subcategory, and a growing set of profiles that apply the framework to a particular technology or setting.

Who it applies to

The AI RMF is written for any organisation that designs, develops, deploys, evaluates or uses AI systems, in the public or private sector and in any country. Nobody is legally bound by it in its own right. It is adopted by choice, or because a customer, investor, regulator or contract asks for AI risk management in its terms.

What it requires

The framework describes outcomes rather than mandatory controls, and the organisation decides the depth that suits its context and risk tolerance.

Trustworthy AI characteristics

NIST describes trustworthy AI as valid and reliable, safe, secure and resilient, accountable and transparent, explainable and interpretable, privacy-enhanced, and fair with harmful bias managed. Validity and reliability are the base the others rest on, and accountability and transparency run across all of them. Risk management is the work of balancing these characteristics for a given system and context.

The four functions

The Core has four functions, 19 categories and 72 subcategories.

• Govern. A culture of risk management, with policies, accountability, roles, workforce diversity and training, engagement with affected people, and processes for third-party and supply chain risk. Govern runs across the other three functions.
• Map. Establish the context of each AI system. Its purpose, intended users and setting, the assumptions it rests on, its capabilities and limits, and the likely impacts on individuals, groups, organisations and society.
• Measure. Analyse, assess, benchmark and monitor the risks that Map identified, using quantitative and qualitative methods, tracking trustworthiness characteristics and collecting feedback over time.
• Manage. Prioritise and respond to risks according to their impact, decide whether a system should proceed, plan responses to incidents and errors, manage third-party risk, and monitor and improve after deployment.

The functions are not sequential. Govern applies throughout, and Map, Measure and Manage repeat as a system changes.

Profiles

Profiles apply the Core to a specific use. The Generative AI Profile (NIST AI 600-1, July 2024) describes 12 risks that are unique to or made worse by generative AI, such as confabulation, information integrity, harmful bias and homogenisation, data privacy, information security, intellectual property and value chain integration. It maps more than 200 suggested actions onto the AI RMF subcategories. NIST released a preliminary draft Cyber AI Profile (NIST IR 8596), built on the Cybersecurity Framework 2.0, in December 2025, and a concept note for a profile on trustworthy AI in critical infrastructure in April 2026.

Key dates

AI RMF 1.0 was released on 26 January 2023 and remains the current version. NIST states that it is being revised as part of the White House AI Action Plan, with no successor published as of October 2026. The Generative AI Profile followed on 26 July 2024.

The framework also carries legal weight in places. Texas’s Responsible Artificial Intelligence Governance Act (HB 149), in force from 1 January 2026, provides that a defendant may not be found liable if it discovers a violation through an internal review process while substantially complying with NIST’s Generative AI Profile or another nationally or internationally recognised AI risk management framework.

How it relates to other frameworks

The AI RMF and ISO/IEC 42001 are complementary. ISO/IEC 42001 is a certifiable management system with clauses an auditor tests. The AI RMF is a risk framework with a deep catalogue of outcomes and actions. Many organisations run the AI RMF’s Map, Measure and Manage work inside a 42001 management system, and NIST publishes crosswalks, tables that map the AI RMF to ISO standards and other frameworks through its Trustworthy and Responsible AI Resource Center.

Its trustworthiness characteristics closely track the OECD AI Principles. The risk management practices it describes map well to the EU AI Act’s risk management, data governance, human oversight and post-market monitoring duties, which makes it a practical working vocabulary for organisations operating across several regimes.

Working with it in Aicura

The NIST AI RMF is available in Aicura from the Pro tier. An administrator enables it in Settings under Frameworks, where it carries the Govern, Map, Measure and Manage structure, and chooses which governance policy reviews apply it under Reviews.

Once enabled, risk and impact assessments for each AI system and agent draw on it alongside your other enabled frameworks, producing rated risk items with written rationale and suggested controls. Those controls are published as a versioned set in Risks and Controls, where each one tracks whether it is applicable, implemented and tested, the kind of record Manage asks for. Results from monitoring, drift checks (which flag when a model’s behaviour shifts over time) and security testing feed the Measure function with live data, and the Policy review report shows where each policy stands against the frameworks it was reviewed under.

Aicura provides guidance for your own risk decisions. It does not certify your work against the AI RMF or vouch for it.