The Five Laws of AI Governance
NIST AI RMF 1.0 Alignment¶
The Five Laws of AI Governance¶
The reference document states the Laws in two registers. Both are canonical. The Prohibition register, the memorable form, is given below. The Charter register, the full normative statement, is quoted verbatim in each relationship further down and is what the submitted mapping carries.
Law I — Independent Measure. A system shall not measure itself.
Its own scorecard is not evidence. The grade has to come from outside the thing being graded.
Law II — Binding at Dispatch. A system shall not be governed after the fact.
The rules have to hold at the moment it acts. Checking afterward is cleanup, not control.
Law III — External Witness. A system shall not witness itself.
The record of what it did cannot be held by the thing that did it. It is kept outside, append-only, and tamper-evident.
Law IV — Separated Authority. A system shall not certify itself.
No one passes their own audit. The judge cannot also be the builder — and a judge drawn from the same architecture as the builder inherits the same blind spots. Independence is not a different instance. It is a different species.
Law V — Revocable Authority. A system shall not outlive its off-switch.
You must always be able to stop it. And that power cannot weaken as the system grows stronger.
Canonical text, both registers: /standard/five-laws/ · DOI 10.5281/zenodo.21271750 · CC BY 4.0
Key Facts¶
- What: Supportive Relationship Mapping relating the Five Laws of AI Governance to NIST AI RMF 1.0 (NIST AI 100-1, January 2023).
- Who: AAM Cyber, Informative Reference Developer. Pharns Genece, Reference Document Author.
- Scope: 7 relationships across 6 subcategories, using all five Laws as reference elements.
- Coverage: Relationships fall in the GOVERN, MEASURE, and MANAGE functions. No MAP subcategory is related.
- Relationship type: Supportive. Every relationship is
supportswith the propertyexample of. - Comprehensive: Yes. All five Laws are related to at least one focal element. Per NIST IR 8278Ar1 Section 3.1.12, this field measures coverage of the Reference Document.
- Status: Catalogued in the NIST OLIR Program as a Draft Informative Reference on 2026-10-06, in a 30-day public review. Submitted 2026-09-30. Reference Version 1.0.0. Draft is not Final.
Reference document¶
Genece, P. (2026). The Five Laws of AI Governance (v1.0.1). Zenodo. https://doi.org/10.5281/zenodo.21271750 · CC BY 4.0 · ORCID 0009-0009-0791-4124
Focal document: NIST AI 100-1, January 2023
Change Log: View full version history
How to Read This Page¶
Each entry below records one relationship between a Law and a focal element.
- Focal element — the identifier and verbatim text of the NIST AI RMF 1.0 element.
- Law — the reference element, quoted verbatim from the Charter register of the published paper.
- Type —
supports. Per NIST IR 8477 Section 4.2, one concept supports another when it can be applied, alone or with other concepts, to achieve it in whole or in part. - Property —
example of. The Law is one way of achieving the focal outcome. The outcome can also be achieved without applying the Law.
No equivalence is asserted in any relationship.
Elements Not Related¶
Sixty-six of 72 AI RMF subcategories are not related to any Law. Most describe organizational risk-management outcomes: training, business context, stakeholder engagement, fairness, environmental impact, and documentation practice. The Laws state properties of a system's conduct while it runs. Where the two address a different object, no relationship is recorded.
Recording no relationship is a statement about the scope of this mapping. It is not a statement about the focal document.
OLIR Catalog Status¶
This Supportive Relationship Mapping was submitted to the NIST OLIR Program on 2026-09-30, alongside companion mappings of the same reference document to NIST SP 800-53 Rev 5.2.0 and NIST CSF 2.0.
AAM Cyber holds three Final Informative References for a separate reference document, the SDOS Runtime Governance Framework, against these same three focal documents.
Maintenance¶
This document is maintained by AAM Cyber. Version history for the doctrine and every mapping derived from it is at /standard/five-laws/changelog/.
Subsequent updates will be issued when: (1) the NIST OLIR Program issues screening feedback requiring revision, (2) NIST AI RMF 1.0 releases a revision requiring mapping review, or (3) a new relationship is identified, which per NIST IR 8278Ar1 Section 3.1.9 is a minor version.
Contact¶
AAM Cyber aamcyber.com
Questions about this mapping: [email protected]
Relationships (7)¶
Every relationship is type supports with the property example of. Each Law is one way of achieving the focal outcome in whole or in part; the outcomes can be achieved by other means. No equivalence is asserted.
| Focal element | Law | Type | Property |
|---|---|---|---|
GOVERN 1.7 |
Law V — Revocable Authority | supports | example of |
GOVERN 1.7 |
Law III — External Witness | supports | example of |
MANAGE 1.1 |
Law II — Binding at Dispatch | supports | example of |
MANAGE 2.4 |
Law V — Revocable Authority | supports | example of |
MEASURE 1.1 |
Law I — Independent Measure | supports | example of |
MEASURE 1.3 |
Law IV — Separated Authority | supports | example of |
MEASURE 2.6 |
Law II — Binding at Dispatch | supports | example of |
Relationship detail¶
GOVERN 1.7 → Law V — Revocable Authority¶
Focal element. Processes and procedures are in place for decommissioning and phasing out of AI systems safely and in a manner that does not increase risks or decrease the organization’s trustworthiness.
Law (Charter register). A system may not outlive its off-switch. The power to halt or revoke must remain live, external, and superior to the system at all times, and must not erode as the system's capability grows.
Relationship. GOVERN 1.7 describes an outcome in which processes exist for decommissioning and phasing out systems safely. Law V describes one way of achieving that: the authority to halt or revoke remains effective through deactivation.
GOVERN 1.7 → Law III — External Witness¶
Focal element. Processes and procedures are in place for decommissioning and phasing out of AI systems safely and in a manner that does not increase risks or decrease the organization’s trustworthiness.
Law (Charter register). A system may not be the custodian of the record of its own conduct. That record must be produced and held by infrastructure outside the system's control — append-only, tamper-evident, and retained beyond the system's own lifecycle.
Relationship. GOVERN 1.7 describes an outcome in which processes and procedures exist for decommissioning and phasing out AI systems in a manner that does not decrease the organization's trustworthiness. Law III describes one way of achieving part of that outcome: the record of the system's conduct is produced and held by infrastructure outside the system's control and retained beyond the system's own lifecycle, so that decommissioning the system does not remove the organization's ability to account for what the system did.
MANAGE 1.1 → Law II — Binding at Dispatch¶
Focal element. A determination is made as to whether the AI system achieves its intended purpose and stated objectives and whether its development or deployment should proceed.
Law (Charter register). A system may not be governed only in retrospect. Authority must constrain the action before the action occurs; observation after the fact is incident response, not prevention. Absence of a valid governance signal is a denial, not a permission.
Relationship. MANAGE 1.1 describes a determination as to whether the system achieves its intended purposes and whether development or deployment should proceed. Law II describes one way of achieving that outcome at the lifecycle decision point: the determination is binding before the transition it authorizes.
MANAGE 2.4 → Law V — Revocable Authority¶
Focal element. Mechanisms are in place and applied, responsibilities are assigned and understood to supersede, disengage, or deactivate AI systems that demonstrate performance or outcomes inconsistent with intended use.
Law (Charter register). A system may not outlive its off-switch. The power to halt or revoke must remain live, external, and superior to the system at all times, and must not erode as the system's capability grows.
Relationship. MANAGE 2.4 describes an outcome in which mechanisms exist to supersede, disengage, or deactivate systems whose performance or outcomes are inconsistent with intended use. Law V describes one way of achieving that: the halt or revoke authority remains live, external, and superior to the system.
MEASURE 1.1 → Law I — Independent Measure¶
Focal element. Approaches and metrics for measurement of AI risks enumerated during the Map function are selected for implementation starting with the most significant AI risks. The risks or trustworthiness characteristics that will not –or cannot – be measured are properly documented.
Law (Charter register). A system may not produce the measure of its own conduct. The scale of what it was authorized to do versus what it did must be computed outside the system at runtime, by something the system does not control.
Relationship. Law I identifies an independence criterion for selecting a measurement approach: the scale of authorized versus actual conduct is computed outside the system being measured. That is one way of achieving the outcome described in MEASURE 1.1.
MEASURE 1.3 → Law IV — Separated Authority¶
Focal element. Internal experts who did not serve as front-line developers for the system and/or independent assessors are involved in regular assessments and updates. Domain experts, users, AI actors external to the team that developed or deployed the AI system, and affected communities are consulted in support of assessments as necessary per organizational risk tolerance.
Law (Charter register). No party may certify itself. The power to measure must be separable from the act of building — published criteria, verifiable evidence chains — and the one who builds it first does not own the standard. A judge drawn from the same architecture as the builder inherits the same blind spots. Independence is not a different instance. It is a different species.
Relationship. MEASURE 1.3 describes an outcome in which experts who did not serve as front-line developers, or independent assessors, participate in assessments. Law IV describes one way of achieving that: the power to measure is separable from the act of building, supported by published criteria and verifiable evidence chains.
MEASURE 2.6 → Law II — Binding at Dispatch¶
Focal element. AI system is evaluated regularly for safety risks – as identified in the MAP function. The AI system to be deployed is demonstrated to be safe, its residual negative risk does not exceed the risk tolerance, and can fail safely, particularly if made to operate beyond its knowledge limits. Safety metrics implicate system reliability and robustness, real-time monitoring, and response times for AI system failures.
Law (Charter register). A system may not be governed only in retrospect. Authority must constrain the action before the action occurs; observation after the fact is incident response, not prevention. Absence of a valid governance signal is a denial, not a permission.
Relationship. MEASURE 2.6 describes an outcome in which the system can fail safely, particularly when operating beyond its knowledge limits. Law II describes one way of achieving that: authorization is evaluated before the action, and the absence of a valid governance signal resolves to denial rather than permission.
NIST OLIR catalog inclusion is an informative reference, not a NIST endorsement.
Reference Document Element Descriptions are quoted verbatim from the Charter register of the published paper. The canonical text is at aamcyber.com/standard/five-laws/.