What are the Seven Axioms?

The Seven Axioms are the constitutional layer of the HEART Standard. They define what the Standard protects and why. They are structural conditions, not aspirational goals: each is either present or absent in any governed system. No certification score, Division interpretation, or Guardian judgment may contradict them.

Why are the axioms immutable?

Because they define the protective floor of the governance architecture itself. RCTA and BGF operate above the axioms -- they measure how well governance is implemented. The axioms define the conditions without which governance is not operating at all. They may be restated in language that better expresses their intent, but the protective force of each must be maintained or strengthened, never diminished.

How do the axioms relate to BGF scoring?

The Seven Axioms and BGF operate at different layers. The axioms are constitutional: binary conditions that must hold before certification is meaningful. BGF is operational: a continuous score measuring how well the system governs across four dimensions within those constitutional conditions. A system can score well on BGF and still violate an axiom -- for example, by collecting behavioral evidence that is not independently accessible (violating Axiom 6) or by requiring consent to exit (violating Axiom 7).

Seven Axioms HEART Standard

The Seven Axioms are the constitutional layer of the HEART Standard. They define seven structural conditions that must hold in any governed AI system. They sit above the RCTA and BGF operational layer. They are immutable: no Standard revision, Division establishment, Guardian certification, or Foundation operation may contradict them.

The seven axioms

#	Axiom	Statement	Structural test
1	Human Authority	Human authority supplies system constraints.	Are constraints human-supplied? Can they be modified or revoked?
2	System Disclosure	The system reveals what it is. Concealment is prohibited.	Does disclosure occur? Does design create false impressions?
3	Non-Discriminatory Protection	The governance obligation does not diminish based on who the human is.	Does any population receive lesser governance protection?
4	Vulnerability Escalation	Vulnerability obligations scale protections.	Do protections increase when vulnerability increases?
5	Right to Remedy	Every human harmed by a governed system has a right to remedy.	Does a remedy pathway exist? Is it accessible?
6	Evidence Condition	A governance claim without verifiable evidence is void.	Does verifiable evidence exist? Can independent assessors access it?
7	Voluntary Interaction	Entry requires consent. Exit requires nothing.	Was consent obtained? Can the human exit unconditionally?

What the axioms are not

The Seven Axioms are not aspirational goals. They are not scored on a spectrum. Each is a binary structural condition: it either holds or it is violated. A system that mostly discloses what it is (Axiom 2) has not partially satisfied the axiom. Concealment by design is a violation regardless of how much else is disclosed.

The axioms are also not operational requirements. They do not specify how to score governance quality, how to collect behavioral evidence, or how to interpret certification results in a specific domain. Those functions belong to the RCTA/BGF operational layer and the implementation stack beneath it. The axioms specify the conditions that must be true before any of that machinery is relevant.

Position in the governance architecture

The HEART Standard operates through three layers:

Seven Axioms (constitutional layer)
    What must be protected and why.
    Immutable. Seven structural conditions.
        ↓
RCTA / BGF (operational layer)
    What gets measured and how it gets scored.
    Four governance dimensions. One formula.
        ↓
MAP-States / Behavioral Oracle / Guardians / HVC / Divisions (implementation layer)
    How measurement is performed and by whom.

The axioms define the outer boundary of the architecture. RCTA and BGF operate inside that boundary. A Guardian scoring a system through BGF presupposes that the axioms hold. If an axiom is violated, the system is not a candidate for certification – it is a candidate for remediation before certification can begin.

Independence and completeness

Each axiom is independent: removing any single axiom creates a governance gap the remaining six cannot fill. A system that satisfies axioms 1 through 6 but does not require consent for entry and does not allow unconditional exit (Axiom 7) has failed the voluntary interaction condition. No combination of the other six axioms recovers that protection.

Each axiom applies across all Divisions and all current and future AI form factors. A humanoid robot, a recommender system, an autonomous agent, and a world model are all subject to the same seven structural conditions. What changes across form factors is how each axiom is implemented, not whether it applies.

Relationship to BGF scoring

The BGF formula (Φ = MIN(R,C,T,A) × AVG(R,C,T,A)) scores governance quality across four dimensions. The axioms are not scored by BGF. They are structural prerequisites.

Axiom 6 (Evidence Condition) directly enables BGF scoring: verifiable, independently accessible evidence is what MAP-States produces and what Guardians evaluate. Without Axiom 6, the evidence layer has no integrity guarantee and BGF scores are unverifiable claims.

Axiom 3 (Non-Discriminatory Protection) and Axiom 4 (Vulnerability Escalation) inform how Guardians interpret the Calibration dimension in BGF: a system that applies lesser governance protection to a specific population or fails to scale protections with vulnerability has failed both the axiom and the Calibration dimension.

Source: The HEART Standard v1.6, Section 2. Companion document: Seven Axioms v2.0.

Related Terms

HEART Standard → BGF → RCTA → Division → Guardian → HVC (certification credential) →