NIST AI RMF in 2026: The Four Functions, the Generative AI Profile (NIST AI 600-1), and the Texas Defense That Names It

What is the NIST AI RMF?

The NIST AI RMF is a free, voluntary framework for managing the risks that AI systems pose to individuals, organizations, and society. NIST released AI RMF 1.0, published as NIST AI 100-1, on January 26, 2023, after a request for information, a concept paper, two public drafts, and workshops. NIST does not certify anyone against it, and the framework states that it “does not prescribe risk tolerance”; each organization decides which risks it will accept.

Part 1 of the document frames AI risk and sets out the characteristics of trustworthy AI: valid and reliable; safe; secure and resilient; accountable and transparent; explainable and interpretable; privacy-enhanced; and fair with harmful bias managed. NIST treats valid and reliable as the necessary base for the others, and accountable and transparent as relating to all of them. Part 2 is the Core, where the four functions live, and a short section on profiles.

What are the four core functions of the NIST AI RMF?

The four functions are GOVERN, MAP, MEASURE, and MANAGE. GOVERN is cross-cutting and informs the other three. NIST says that after the GOVERN outcomes are in place, most users start with MAP and continue to MEASURE or MANAGE, and that the actions “do not constitute a checklist.”

The four AI RMF 1.0 functions

Function	What it asks	Example categories	Evidence to keep
GOVERN	A culture of risk management, with policies, accountability, and processes across the organization, including for third-party software and data	GOVERN 1 (policies and practices); GOVERN 2 (accountability structures); GOVERN 6 (third-party and supply chain)	AI policy, role assignments, training records, the AI system inventory (GOVERN 1.6), decommissioning procedure (GOVERN 1.7), vendor terms
MAP	The context of each system: intended purpose, users, settings, applicable laws, components, and who could be affected	MAP 1 (context); MAP 4 (risks and benefits of all components); MAP 5 (impacts)	Intended-use statement (MAP 1.1), system categorization, component and third-party risk list, impact analysis
MEASURE	Testing and metrics for the risks MAP found, before deployment and regularly in operation, with tracking over time	MEASURE 1 (methods and metrics); MEASURE 2 (evaluation for trustworthy characteristics); MEASURE 3 (risk tracking)	Test sets, metrics, and tools used (MEASURE 2.1), security results (MEASURE 2.7), fairness and bias results (MEASURE 2.11), list of risks that cannot be measured (MEASURE 1.1)
MANAGE	Prioritizing and treating risks, deciding whether to proceed, monitoring after deployment, and responding to incidents	MANAGE 1 (prioritization and response); MANAGE 2 (strategies to maximize benefits and minimize impacts); MANAGE 4 (treatment, response, recovery, communication)	Go or no-go decision (MANAGE 1.1), shutdown mechanism and owner (MANAGE 2.4), post-deployment monitoring plan (MANAGE 4.1), incident log (MANAGE 4.3)

What are the 19 categories of the AI RMF Core?

Each function is divided into numbered categories, and each category into subcategories. The 19 categories split six under GOVERN, five under MAP, and four each under MEASURE and MANAGE. The 72 subcategories split 19, 18, 22, and 13 in the same order. The category titles below are NIST’s own wording from Tables 1 to 4 of AI 100-1.

AI RMF 1.0 categories, as titled in NIST AI 100-1

ID	Category
GOVERN 1	Policies, processes, procedures, and practices across the organization related to the mapping, measuring, and managing of AI risks are in place, transparent, and implemented effectively.
GOVERN 2	Accountability structures are in place so that the appropriate teams and individuals are empowered, responsible, and trained for mapping, measuring, and managing AI risks.
GOVERN 3	Workforce diversity, equity, inclusion, and accessibility processes are prioritized in the mapping, measuring, and managing of AI risks throughout the lifecycle.
GOVERN 4	Organizational teams are committed to a culture that considers and communicates AI risk.
GOVERN 5	Processes are in place for robust engagement with relevant AI actors.
GOVERN 6	Policies and procedures are in place to address AI risks and benefits arising from third-party software and data and other supply chain issues.
MAP 1	Context is established and understood.
MAP 2	Categorization of the AI system is performed.
MAP 3	AI capabilities, targeted usage, goals, and expected benefits and costs compared with appropriate benchmarks are understood.
MAP 4	Risks and benefits are mapped for all components of the AI system including third-party software and data.
MAP 5	Impacts to individuals, groups, communities, organizations, and society are characterized.
MEASURE 1	Appropriate methods and metrics are identified and applied.
MEASURE 2	AI systems are evaluated for trustworthy characteristics.
MEASURE 3	Mechanisms for tracking identified AI risks over time are in place.
MEASURE 4	Feedback about efficacy of measurement is gathered and assessed.
MANAGE 1	AI risks based on assessments and other analytical output from the MAP and MEASURE functions are prioritized, responded to, and managed.
MANAGE 2	Strategies to maximize AI benefits and minimize negative impacts are planned, prepared, implemented, documented, and informed by input from relevant AI actors.
MANAGE 3	AI risks and benefits from third-party entities are managed.
MANAGE 4	Risk treatments, including response and recovery, and communication plans for the identified and measured AI risks are documented and monitored regularly.

What are AI RMF profiles and the Playbook?

A profile is an implementation of the RMF’s functions, categories, and subcategories for a specific setting, application, or technology. AI 100-1 describes three kinds. Use-case profiles fit a specific application; NIST gives a hiring profile and a fair housing profile as examples. Temporal profiles compare a Current Profile, how AI risk is managed today, with a Target Profile, the outcomes needed, so the gap becomes an action plan. Cross-sectoral profiles cover risks common across sectors, such as large language models, cloud-based services, or acquisition. The framework prescribes no profile template.

The Playbook is NIST’s online companion to the RMF. It offers suggested actions, references, and related guidance for the outcomes in the four functions, and it is voluntary: an organization may borrow as many or as few suggestions as fit. NIST states that the Playbook will be updated after AI RMF 1.0 is revised.

Besides the Generative AI Profile, covered below, NIST’s AI RMF page lists one profile in development. On April 7, 2026, NIST released a concept note for an AI RMF Profile on Trustworthy AI in Critical Infrastructure, meant to guide critical infrastructure operators toward specific risk management practices. What NIST released is a concept note, not the profile.

What does the NIST Generative AI Profile (NIST AI 600-1) add?

NIST AI 600-1, released July 26, 2024, is a cross-sectoral profile of the RMF for generative AI. It defines 12 risks that are unique to or made worse by generative AI, then lists suggested actions to govern, map, measure, and manage them. The 12 risks, in NIST’s order and wording:

1. CBRN Information or Capabilities
2. Confabulation
3. Dangerous, Violent, or Hateful Content
4. Data Privacy
5. Environmental Impacts
6. Harmful Bias or Homogenization
7. Human-AI Configuration
8. Information Integrity
9. Information Security
10. Intellectual Property
11. Obscene, Degrading, and/or Abusive Content
12. Value Chain and Component Integration

The actions are organized by RMF subcategory. Each action ID names the function and subcategory it serves: GV-1.1-001 is the first suggested action for GOVERN 1.1, and the prefixes are GV for Govern, MP for Map, MS for Measure, and MG for Manage. Each action is tagged with the generative AI risks it addresses. GV-1.1-001, for example, asks an organization to align generative AI development and use with applicable laws and regulations and is tagged Data Privacy, Harmful Bias and Homogenization, and Intellectual Property.

Coverage is uneven. Not every subcategory has actions, and NIST notes that actions relevant to developers may not be relevant to deployers. An appendix adds four primary considerations: governance, pre-deployment testing, content provenance, and incident disclosure. AI 600-1 also states that the activities in AI RMF 1.0 and the Playbook already apply to generative AI, so the profile supplements the framework rather than replacing it. For a Texas defendant, though, the profile is the document the statute names.

Is NIST revising the AI RMF?

Yes, but no revised text had been published as of November 2, 2026. The July 23, 2025 White House AI Action Plan directed NIST to revise the RMF to remove references to misinformation, diversity, equity, and inclusion, and climate change. NIST’s AI RMF page states that “The AI RMF 1.0 is being revised as part of the White House AI Action Plan,” and the NIST AI Resource Center states that a revised version is in progress. Neither page links to a draft. A Federal Register search found no NIST request for information or comment on a revision; the framework’s own Federal Register notices date from 2021.

The directive names AI 100-1; it does not name AI 600-1. NIST has not said which categories will change. The closest matches to the directive’s topics in the current text are these. In the 1.0 text, GOVERN 3 is the category whose title names workforce diversity, equity, inclusion, and accessibility. In AI 600-1, Information Integrity covers dis- and misinformation campaigns, and Environmental Impacts covers high compute resource use.

Vocabulary may change too. Executive Order 14434, signed September 29, 2026, directs executive departments and agencies to use “Super Intelligence” and “SI” in place of “Artificial Intelligence” and “AI” in their own non-statutory documents. It binds those departments and agencies only; a company’s own documents, and statutes such as Texas HB 149 that cite the NIST profile by its title, are unaffected. NIST has begun renaming projects under that order. On its control overlays project page it states that previously issued documents and publications have not been altered, and AI 100-1 and AI 600-1 still carried their original titles on NIST’s site on November 2, 2026.

Does CAISI run the NIST AI RMF, and which other NIST documents matter?

No. NIST’s RMF page says the framework is led by the Information Technology Laboratory AI program. The Center for AI Standards and Innovation (CAISI), housed within NIST, is the government’s point of contact for industry on testing and collaborative research. By October 3, 2026, NIST’s site had renamed it the Center for Advancing Innovation and Standards for Super Intelligence (CAISSI); we found no formal rename notice. It develops guidelines and best practices for measuring and improving the security of AI systems, makes voluntary agreements with private developers and evaluators, and leads unclassified evaluations of capabilities that may pose national security risks, with a focus on demonstrable risks such as cybersecurity, biosecurity, and chemical weapons. Its January 8, 2026 request for information on AI agent security cites the RMF’s security subcategories in MEASURE as existing NIST material. On October 16, 2026, it published ten voluntary key practices for incident readiness and response in training and evaluation with the NSA and CISA.

Three related NIST documents fill gaps the RMF leaves on security:

• NIST AI 100-2 E2025, Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations, final March 24, 2025. It names attack types and mitigations; use it when MEASURE 2.7 security testing needs a common vocabulary.
• SP 800-53 control overlays. NIST published a concept paper on August 14, 2025 and an annotated outline for a predictive AI overlay on January 8, 2026. On October 20, 2026 the project, previously Control Overlays for Securing AI Systems (COSAiS), was renamed Control Overlays for Super Intelligence Systems (COSIS), citing Executive Order 14434. As of November 2, 2026 the project lists no published overlay.
• NIST IR 8596, Cyber AI Profile, a preliminary draft of December 16, 2025 organized on Cybersecurity Framework 2.0 outcomes; comments closed January 30, 2026.

Where does the law give the NIST AI RMF weight?

Among the laws we checked, two state statutes name it directly: Texas HB 149, in a liability defense, and Montana’s Right to Compute Act, in a risk management policy duty for AI-controlled critical infrastructure that any of three frameworks can satisfy. Colorado’s 2026 replacement law dropped the 2024 framework defense. California SB 53, the 2025 OMB memoranda, and the two New York DFS letters do not name it. State insurance bulletins based on the NAIC model name it as one framework an insurer’s AI program may rely on, without requiring it. The two federal documents below that do name it, a GSA draft contract clause and an FDA request for information, impose no duties.

Where the NIST AI RMF appears in the laws and agency documents these guides cover, as of November 2, 2026

Instrument	Names the RMF or AI 600-1?	Effect
Texas HB 149, Bus. & Com. Code Sec. 552.105(e)	Yes, the Generative AI Profile	No liability for a violation discovered through internal review while substantially complying with the most recent version of the profile or another recognized framework
Montana Right to Compute Act, Mont. Code Ann. 2-10-205 (2025)	Yes, the latest version of the AI RMF	When critical infrastructure facilities are controlled in whole or in part by a critical artificial intelligence system, the deployer must develop, after deploying the system, a reasonable risk management policy that considers guidance and standards in the latest version of the NIST AI RMF, the ISO/IEC AI standard, or another nationally or internationally recognized AI risk management framework; a plan prepared under federal requirements constitutes compliance
Colorado SB 26-189	No; the repealed 2024 law named it	No framework defense
California SB 53, Bus. & Prof. Code 22757.12	No	A large frontier developer’s published framework must describe how the developer approaches incorporating national standards, international standards, and industry-consensus best practices into it
OMB M-25-21 and M-25-22	No	Seven minimum practices for federal agencies’ high-impact AI; binds agencies only
GSA draft GSAR clause, 91 FR 36559 (June 17, 2026)	Yes, AI RMF 1.0 Appendix A and NIST AI RMF guidelines	Draft only; role definitions for LLM developers, providers, integrators, and operators, and, on government request, system documentation consistent with NIST AI RMF guidelines
NYDFS Insurance Circular Letter No. 7 (2024)	No	Expects a governance and risk management framework without naming one
NAIC Model Bulletin on the Use of AI Systems by Insurers (December 4, 2023), as adopted by state insurance departments, such as Connecticut Bulletin MC-25	Yes, AI RMF 1.0, as an example	An insurer’s written AIS Program may adopt, incorporate, or rely upon, in whole or in part, a framework such as the RMF; not required

Texas. Under Sec. 552.105(e), a defendant may not be found liable if another person used the defendant’s AI system in a prohibited manner, or if the defendant discovered the violation through feedback from a developer, deployer, or other person; through testing, including adversarial or red-team testing; by following guidelines set by applicable state agencies; or through an internal review process, if the defendant substantially complies with the most recent version of the NIST Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile or another nationally or internationally recognized AI risk management framework. The framework clause attaches to the internal review route. A company that adopts the NIST profile but never runs the internal review that finds the problem gets nothing from the framework alone. Because the statute points to the most recent version, a revised profile, once NIST publishes one, is the text to watch. The Texas HB 149 guide covers the rest of the Act.

Montana. Under Montana’s Right to Compute Act, enacted in 2025, when critical infrastructure facilities are controlled in whole or in part by a critical artificial intelligence system, the deployer must develop a risk management policy after deploying the system that is reasonable and considers guidance and standards in the latest version of the NIST AI RMF, the ISO/IEC artificial intelligence standard from the International Organization for Standardization, or another nationally or internationally recognized risk management framework for AI systems. A plan prepared under federal requirements constitutes compliance (Mont. Code Ann. 2-10-205). The duty is to consider the RMF, not to adopt it, and any of the three frameworks can satisfy it; the code prints the ISO/IEC number as “4200”. A deployer is any individual, company, or other organization that uses an AI system. A critical artificial intelligence system is one designed and deployed to make, or to be a substantial factor in making, a consequential decision, with listed exclusions, and “critical infrastructure facility” takes its meaning from Mont. Code Ann. 82-1-601 (Mont. Code Ann. 2-10-203).

Colorado. SB 26-189 keeps notice, adverse-outcome explanation, data correction, and human review. It drops the 2024 Act’s risk management program, impact assessments, public statements, reports to the Attorney General, and the NIST and ISO affirmative defense. A federal court order in X.AI LLC v. Weiser bars the Attorney General from initiating enforcement, including an investigation, for violations occurring on or before 14 days after the court rules on xAI’s preliminary injunction motion. As of November 2, 2026, the CourtListener docket showed no such motion; its last entry was dated October 2, 2026. See the Colorado SB 26-189 guide.

California. A large frontier developer must write, implement, comply with, and publish a frontier AI framework covering, among other things, catastrophic risk thresholds, mitigations, third-party assessments, cybersecurity of unreleased weights, and internal governance, and may not make a materially false or misleading statement about its implementation of, or compliance with, that framework. The first item the framework must describe is how the developer approaches incorporating national standards, international standards, and industry-consensus best practices (Sec. 22757.12(a)(1)). The statute does not name NIST. A developer that names the RMF in its framework should be able to show it follows it, because the statute bars materially false or misleading statements about compliance with the framework. See the California AI laws guide.

Federal agencies. Executive Order 14110, revoked in January 2025, had directed OMB to set minimum risk-management practices for federal agencies’ AI, including, where appropriate, practices derived from the RMF and the 2022 Blueprint for an AI Bill of Rights. M-25-21 states that it “rescinds and replaces” M-24-10, and M-25-22 states that it rescinds and replaces M-24-18. Neither 2025 memorandum mentions NIST or the RMF. For high-impact AI, agencies must apply seven minimum practices: pre-deployment testing, an AI impact assessment with an independent reviewer, ongoing monitoring for performance and adverse impacts, operator training, human oversight with a fail-safe where practicable, “timely human review and a chance to appeal” for people affected by AI-enabled decisions, and a channel for feedback from end users and the public. A vendor selling AI to federal agencies should expect agencies working under M-25-21 and M-25-22 to ask for testing, impact assessment, and monitoring information; an RMF program is one way to have it ready.

GSA’s June 17, 2026 draft contract clause on safeguarding data in large language model systems maps its role definitions to the AI actor tasks in AI RMF 1.0, Appendix A, and would require the contractor to provide, on government request, system documentation consistent with NIST AI Risk Management Framework guidelines; comments were due August 3, 2026, and we found no final clause. An April 29, 2026 FDA request for information states that its AI clinical trials pilot will be guided by principles aligned with the RMF; a request for information imposes no duties.

New York DFS. Neither NYDFS Insurance Circular Letter No. 7 (July 11, 2024) nor the department’s October 16, 2024 industry letter on AI cybersecurity risks names the RMF. The October letter’s footnotes link to other NIST material: a glossary entry for defense-in-depth and NIST’s zero trust architecture publication.

State insurance bulletins. The NAIC adopted its Model Bulletin on the Use of Artificial Intelligence Systems by Insurers on December 4, 2023; as of October 8, 2026, the NAIC listed 27 states, the District of Columbia, and Puerto Rico as having adopted it, some with changes. In the adopted text, an insurer’s written AI Systems (AIS) Program may adopt, incorporate, or rely upon, in whole or in part, a framework or standards developed by an official third-party standard organization, such as the NIST AI RMF, Version 1.0 (Connecticut Bulletin MC-25, Sec. 1.5). The RMF is an example an insurer may use, not a requirement. The AI compliance in insurance guide covers the bulletin.

How does the NIST AI RMF compare with ISO/IEC 42001?

The RMF is a free framework that a company applies and describes on its own; NIST audits no one against it. ISO/IEC 42001:2023 is a paid management system standard that an independent certification body can audit and certify. Texas names only the NIST profile, but its clause also covers “another nationally or internationally recognized” framework, the wording an ISO/IEC 42001 program would rely on. The AI governance frameworks guide compares the two side by side and explains why it recommends that most companies use both, in sequence: NIST to decide what to do, and ISO/IEC 42001 when someone needs outside proof.

How should a company start using the NIST AI RMF?

• Name an executive owner and write a short AI policy that assigns roles (GOVERN 1, GOVERN 2)
• Build the inventory GOVERN 1.6 calls for: each system, its owner, vendor, purpose, and the decision it touches
• For each system, document intended purpose, users, settings, and applicable laws (MAP 1.1), and record a go or no-go decision (MANAGE 1.1)
• Write a Current Profile and a Target Profile, list the gaps, and date each fix
• For generative AI, pick the AI 600-1 risks that apply and record which suggested actions you adopted by action ID
• Test before deployment and on a schedule, keeping test sets, metrics, and tools (MEASURE 2.1), security results (MEASURE 2.7), and fairness and bias results (MEASURE 2.11)
• Assign who can supersede, disengage, or deactivate a system (MANAGE 2.4), and write the post-deployment monitoring and incident plan (MANAGE 4.1, MANAGE 4.3)
• Put third-party models and data under contract terms and monitoring (GOVERN 6, MANAGE 3)
• If Texas HB 149 reaches you (it applies to a person who promotes, advertises, or conducts business in Texas, produces a product or service used by Texas residents, or develops or deploys an AI system there), substantially comply with the most recent version of the NIST Generative AI Profile or another nationally or internationally recognized AI risk management framework, run the internal review under it on a schedule, and log what it finds and when, so the Sec. 552.105(e) internal review route is available
• If you deploy a critical artificial intelligence system that controls, in whole or in part, a critical infrastructure facility in Montana, develop the reasonable risk management policy Mont. Code Ann. 2-10-205 requires after deployment, and record how it considers the latest NIST AI RMF, the ISO/IEC AI standard, or another recognized framework, or keep the plan prepared under federal requirements that counts as compliance
• Watch NIST’s AI RMF page for the revision, then re-map your profile and the Playbook actions you rely on

How this fits the other guides: the governance frameworks guide is the broader page this one sits under. The AI compliance guide sets out the first 90 days, the AI Bill of Rights guide covers the OMB memoranda, and the agentic AI governance guide covers NIST’s agent work.

This guide is general information as of November 2, 2026, not legal advice. Adopting the NIST AI RMF does not by itself establish compliance with any law; confirm current text with NIST, the relevant legislature, or counsel before acting on it.

By industry: all 8 industry pages