Skip to main content

Documentation Index

Fetch the complete documentation index at: https://verdictweight.dev/llms.txt

Use this file to discover all available pages before exploring further.

What ISO/IEC 42001 is

ISO/IEC 42001:2023 is the international standard for AI management systems (AIMS), published December 2023. It is structurally analogous to ISO/IEC 27001 (information security management) but specific to AI — specifying requirements for establishing, implementing, maintaining, and continually improving an AI management system within an organization. ISO/IEC 42001 is the international certification track for organizations that want a recognized, auditable management system for their AI activities. It complements rather than replaces sector-specific regulation: an organization can be ISO/IEC 42001 certified and still need to satisfy NIST AI RMF, EU AI Act, or DoD-specific requirements.

Coverage summary

ISO/IEC 42001 is a management system standard. The vast majority of its requirements are organizational: policies, processes, leadership commitment, internal audit, continual improvement. VERDICT WEIGHT, as a technical building block, addresses a subset of the standard’s controls — primarily those in Annex A (the reference list of AI controls) and the technical aspects of Clause 8 (operation). The framework supports certification; it does not provide certification.

Annex A controls supported

Annex A of ISO/IEC 42001 enumerates a structured list of AI controls organized into nine categories. VERDICT WEIGHT’s contribution by category: The framework provides artifacts that support policy implementation but does not itself constitute policy.
ControlFramework contribution
A.2.2 – AI policyConfigurable thresholds and registry-protected configuration enforce policy at the technical layer.
A.2.4 – review of AI policyAudit-chain replay supports policy-effectiveness review.

A.4: Resources for AI systems

Resources include data, tooling, computing, and human resources. Framework contribution is primarily in tooling provenance.
ControlFramework contribution
A.4.5 – tooling resourcesThree-source integrity verification (PyPI / GitHub / Zenodo) establishes tooling provenance.
A.4.6 – system and computing resourcesDocumented complexity profile supports resource planning.

A.5: Assessing impacts of AI systems

Impact assessment is largely operator-supplied; the framework provides per-stream interpretability that supports it.
ControlFramework contribution
A.5.2 – AI system impact assessmentPer-stream contributions and audit-chain replay support per-decision impact analysis.
A.5.4 – assessing AI system impacts on individualsAudit-chain replay enables per-individual decision review where data permits.

A.6: AI system life cycle

This is where VERDICT WEIGHT does substantial work. Life-cycle controls map directly to framework primitives.
ControlFramework contribution
A.6.1.2 – objectives for AI system developmentThe threat model, failure taxonomy, and validation criteria are objective statements made auditable.
A.6.1.3 – processes for the responsible design of AI systemsThe framework’s design principles (Architecture overview) document the responsible-design process.
A.6.2.2 – AI system requirements and specificationDocumented threat model and stream specifications.
A.6.2.3 – documentation of AI system designComprehensive technical documentation site.
A.6.2.4 – AI system verification and validation673-test suite + IEEE-grade validation procedure.
A.6.2.5 – AI system deploymentOperator runbooks and pilot engagement procedure.
A.6.2.6 – AI system operation and monitoringAudit chain provides continuous operational evidence.
A.6.2.7 – AI system technical documentationThis documentation site.
A.6.2.8 – AI system event loggingStream 7 cryptographic audit chain.

A.7: Data for AI systems

ControlFramework contribution
A.7.2 – data for development and enhancementValidation dataset documented and reproducible.
A.7.4 – quality of data for AI systemsStream 1 evidence aggregation and Stream 4 cross-source coherence provide data-quality signals.
A.7.5 – data provenanceAudit chain records data references; field hashing supports privacy-preserving provenance.

A.8: Information for interested parties of AI systems

ControlFramework contribution
A.8.2 – system documentationDocumentation site published openly.
A.8.3 – external reportingReproducibility pipeline supports external reporting.
A.8.4 – communication of incidentsAudit-chain event format suitable for incident reporting integration.
A.8.5 – information for interested partiesPublic papers, public source, public validation.

A.9: Use of AI systems

ControlFramework contribution
A.9.2 – processes for responsible use of AI systemsOperator runbook templates in pilot deliverables.
A.9.3 – objectives for responsible use of AI systemsCalibrated confidence + threshold + abstention provides the substrate for objective definition.
A.9.4 – intended useThreat model documents intended use envelope.

A.10: Third-party and customer relationships

Largely operator-managed. Framework contribution is in the IP and licensing posture.
ControlFramework contribution
A.10.2 – allocating responsibilitiesIP posture (USPTO patent + trademark, published source) makes responsibility allocation tractable.
A.10.4 – suppliersThree-source integrity check provides supply-chain evidence.

Clauses 4-10: Management system requirements

The numbered clauses of ISO/IEC 42001 (Context, Leadership, Planning, Support, Operation, Performance evaluation, Improvement) are organizational requirements. VERDICT WEIGHT does not satisfy these directly; it supplies artifacts that support an organization’s satisfaction of them.
ClauseFramework contribution
4 (Context)Threat model documents external and internal context for the framework’s use.
5 (Leadership)Operator-supplied.
6 (Planning)Documented threat model and risk taxonomy support risk-based planning.
7 (Support)Documentation site, published source, reproducibility pipeline support competence and awareness.
8 (Operation)The framework is the operational substrate. Audit chain provides operational evidence.
9 (Performance evaluation)Calibration metrics, audit-chain replay, and benchmark reproducibility support evaluation.
10 (Improvement)Documented refit procedures and version control support continual improvement.

Audit artifacts produced

For an ISO/IEC 42001 internal or external audit, the framework provides:
ArtifactMaps to
Hash-chained audit logA.6.2.8, A.7.5, A.8.4
Test suite results (673/673)A.6.2.4
Validation reproducibility pipelineA.6.2.4, A.8.3
Threat model documentationA.5.2, A.6.2.2
Documentation siteA.6.2.3, A.6.2.7, A.8.2, A.8.5
Three-source integrity verificationA.4.5, A.10.4
Per-stream interpretability dataA.5.2, A.5.4, A.9
Kill-switch event logA.6.2.6, A.8.4

What the operator still owns

ISO/IEC 42001 certification is fundamentally an organizational achievement. The framework cannot supply:
  • Leadership commitment — documented top-management support for the AIMS.
  • Roles and responsibilities — an organizational chart of AI governance.
  • Risk-management policy — a written statement of AI risk appetite.
  • Internal audit programs — the certified internal-audit function.
  • Management review — the recurring management-review process.
  • Continual improvement procedures — the documented improvement workflow.
  • Competence and awareness training — workforce training programs.
These are AIMS-level requirements that any technical building block presupposes rather than provides.

Path to certification

Organizations pursuing ISO/IEC 42001 certification with VERDICT WEIGHT in the deployment scope should:
1

Scope the AIMS

Define the AI systems within scope, including those that use VERDICT WEIGHT as a confidence layer.
2

Map controls

Use the Annex A mapping above to identify which framework artifacts satisfy which controls. Document the mapping.
3

Establish organizational requirements

Build the AIMS clauses 4-10 around the technical substrate. This is the bulk of the certification work.
4

Internal audit

Conduct internal audit. Framework artifacts (audit chain, test results, validation reproducibility) are concrete evidence.
5

External certification

Engage an accredited certification body. Framework documentation supports the technical-readiness portion.

Composability with other regimes

A deployment that satisfies the relevant Annex A controls through VERDICT WEIGHT is well-positioned for: The same audit-chain records, the same calibration evidence, the same test results function as evidence across multiple regimes. This is the composability point made in Compliance overview.