# Automation Trust Inheritance Operates Outside Verifiable Governance Until an Independent Validation Surface Exists at the Point of Execution - **Artifact ID:** CHQ-ASC-2026-004 - **Version:** v1.0 - **Status:** Canonical - **Public record:** https://record.cybersecurityhq.com/constraints/chq-asc-2026-004 - **Machine-record SHA-256:** `5e28f169f52cc49a43908a11b8a3f2c16d8db492dae769ca278f1269e13e95f6` ## Complete structured record ```json { "id": "CHQ-ASC-2026-004", "legacy_ids": [ "CHQ-ACJ-2026-004" ], "title": "Automation Trust Inheritance Operates Outside Verifiable Governance Until an Independent Validation Surface Exists at the Point of Execution", "version": "v1.0", "issued": "2026-03-15", "status": "Canonical", "doctrine_class": "Anticipatory Structural Constraint", "precedent": [ "CHQ-ASC-2026-001", "CHQ-ASC-2026-003" ], "purpose": "This Anticipatory Structural Constraint establishes the governing principle for evaluating governance claims made about automated systems, AI agents, and non-human identities that initiate actions by inheriting trust from their authorization context.\n\nEnterprises increasingly deploy automated systems that are granted execution authority derived from the permissions of the human or system that provisioned them. This constraint defines the interpretive boundary for governance assertions about such systems across all CHQ governance artifacts.", "core_judgment": "Automated systems that inherit trust from their authorization context and initiate state-changing actions without passing through an independent validation surface are operating outside verifiable governance regardless of the policy assertions of the systems that spawned them.\n\nInherited trust is not verified trust. The governance posture of the provisioning system does not transfer to the provisioned agent. Post-execution logging, periodic review, and reliance on the agent's own guardrails are observation mechanisms, not governance. Governance requires an enforcement boundary that is architecturally distinct from the execution path it governs.", "derivation_intro": "This constraint derives from prior canonical constraints and observed structural conditions:", "derivation": [ "From CHQ-ASC-2026-001: Automated systems that initiate system-state changes are classified by their capability, not their intended function or design intent", "From CHQ-ASC-2026-003: Trust state is multi-dimensional and time-dependent; boolean trust signals cannot represent institutional trust state", "Automated systems are provisioned with permissions that reflect the authorization context at time of creation, not at time of execution", "Execution authority inherited at provisioning persists and accumulates independently of subsequent changes to governance intent", "No mechanism currently deployed at enterprise scale verifies that inherited permissions remain appropriate at each point of execution", "Post-execution observation of agent actions does not constitute an enforcement boundary; it documents what has already occurred", "Systems relying on inherited trust without independent pre-execution validation cannot produce governance claims that survive adversarial review" ], "scope_governs": [ "Governance assertions about AI agents granted execution authority over production systems", "Governance assertions about automated workflows that initiate state-changing actions", "Interpretation of non-human identity permissions and their inherited authorization scope", "Any CHQ Position, analysis, or assessment evaluating the governance posture of automated or agentic systems", "Evaluation of human-in-the-loop control claims where no independent validation surface exists at the execution boundary" ], "scope_does_not_govern": [ "The design of specific agent architectures or validation mechanisms", "Selection of identity governance or PAM technologies", "Operational response procedures for agent incidents", "Evaluation of agent performance, accuracy, or reliability" ], "relationship_to_derived": "CHQ Positions applying this constraint to specific architectures, technologies, or deployment contexts inherit its authority but do not inherit its permanence. Domain-specific conclusions may change as architectures evolve. This constraint does not.\n\nCHQ-P-2026-005 (AI Agent Execution Authority Requires Independent Deterministic Validation) derives directly from this constraint and applies it to the specific governance condition of pre-execution validation in enterprise AI deployments.", "irreversibility_statement": "This constraint contains no retirement triggers.\n\nIt may be superseded only by a successor Anticipatory Structural Constraint demonstrating that inherited trust in automated systems can be verified as appropriate at the point of execution without an architecturally independent validation surface. Deployment velocity, operational convenience, or the absence of confirmed incidents are not sufficient grounds for supersession.", "boundary_of_application": [ "Invalidate specific products, vendors, or agent frameworks", "Assign liability for governance failures in automated systems", "Prescribe specific validation architectures or controls", "Replace regulatory or legal compliance requirements", "Address the performance, safety, or alignment properties of AI systems" ], "attestation": "This constraint reflects CHQ's position that inherited trust is not verified trust, and that governance assertions about automated systems that lack an independent validation surface at the point of execution cannot be structurally substantiated. Security governance must treat the absence of such a surface as a governance gap regardless of the policy posture of the provisioning system.", "pdf_hash": "a6ea5cdc2bda082f5296df64ec83928b1c1f08169a7b36d419d612d9b9c47bdc", "derived_artifacts": [ "CHQ-P-2026-005", "CHQ-P-2026-017" ], "classification_scope": "EXTERNAL_INTERPRETIVE", "classification_resolution": "This constraint governs interpretation of an external structural condition.", "lifecycle_governance": { "error_correction": "A clerical, citation, date, identifier, or transcription error may be corrected only by a dated erratum that identifies the prior text, corrected text, evidence for the correction, affected hash, and whether meaning changed. Meaning-changing corrections are amendments, not errata.", "amendment": "A constraint may be amended only by a ratified, versioned amendment that preserves the prior version, states the changed proposition and rationale, recomputes the canonical hash, and notifies registered reliance. An amendment may clarify or narrow a constraint but may not silently replace its core judgment.", "successor": "A constraint may be superseded only by a separately identified successor that cites the predecessor, satisfies the predecessor's stated supersession test, states the exact proposition displaced, and records the effective transition. Prior versions remain historical and ineligible for new reliance after supersession.", "permanence": "Permanent means no automatic expiry or retirement trigger. It does not mean immune from demonstrated error, transparent amendment, or a successor that satisfies the canonical supersession test." }, "governance_amendments": [ { "date": "2026-09-22", "effective_version": "v1.0", "amendment_id": "CHQ-ASC-TEMPLATE-2026-001", "description": "Successor and reassessment language normalized as explicit artifact-governance metadata; individual canonical supersession tests remain controlling.", "canonical_hash_effect": "UNCHANGED" }, { "date": "2026-09-22", "effective_version": "v1.0", "amendment_id": "CHQ-PRS-2026-001-AMD-003", "description": "Class-wide error-correction, amendment, successor, and permanence semantics ratified as non-canonical governance metadata. ASC-006's internal-adjudication classification is formally resolved without changing its canonical constraint text.", "canonical_hash_effect": "UNCHANGED" } ] } ```