About This Document This is a reference example showing how certification bodies can incorporate SWT3 witness evidence into ISO/IEC 42001 Stage 2 certification audits. All organization names, personnel, and anchor fingerprints are fictional. Adapt this format to your certification body's QMS and reporting requirements.
This Is Not a Real Assessment This document is illustrative. It does not certify, approve, or endorse any product or system. It demonstrates a citation format, not a certification decision.

Report Cover Page

Stage 2 Certification Audit Report

ISO/IEC 42001:2023

Artificial Intelligence Management System (AIMS)

Organization Audited: Verdant Health Technologies AG

AI System in Scope: ClinAssist v2.0 (Clinical Pathway Recommendation Engine)

AIMS Scope: Development and deployment of clinical decision support AI systems

Certification Body: International AI Certification Services Ltd. (IACS)

CB Accreditation: CB-4192 (Fictional)

Audit Standard: ISO/IEC 42001:2023

Audit Type: Stage 2 (Initial Certification)

On-site Audit: 2026-09-24 through 2026-09-26 (3 days)

Evidence Review Period: 2026-06-01 through 2026-09-30 (122 days)

Total Audit Time: 8 person-days (2 auditors)

Report Date: 2026-10-08

1. Executive Summary

Scope of Audit

This Stage 2 certification audit evaluates Verdant Health Technologies' AI Management System (AIMS) against ISO/IEC 42001:2023 requirements (Clauses 4-10) and applicable Annex A controls. ClinAssist v2.0 is a clinical pathway recommendation engine deployed across 8 hospital networks in Germany and Switzerland, providing treatment pathway suggestions to clinical staff.

Evidence Sources

  1. Cryptographic witness evidence (SWT3): 25 SWT3 witness anchors covering 16 procedures, assessed via the audit portal over the 122-day evidence review period. Each anchor was independently verified.
  2. Document review: AI management system manual, AI policy, Statement of Applicability (SoA), risk assessment methodology, data management procedures, supplier evaluation records, and management review minutes.
  3. Interviews: 5 sessions with the AIMS Manager, Chief Medical Officer, Data Protection Officer, AI Development Lead, and Head of Clinical Operations. Interview notes on file (ref: IACS-INT-4192-01 through 05).
  4. On-site observation: Clinical deployment environment, model monitoring dashboards, and incident response procedures observed during 3-day on-site audit (2026-09-24 through 2026-09-26).

Audit Outcome

CONDITIONAL CERTIFICATION RECOMMENDED. 6 of 7 clause groups are conforming. One minor nonconformity identified in Clause 9 (management review overdue). Two observations recorded. Annex A: 28 of 31 applicable controls conforming.

MetricValue
Procedures Assessed16
Conforming13
Minor Nonconformity1
Observations2
Annex A Controls (Applicable / Conforming)31 / 28
SWT3 Anchors Verified25 (5 independently recomputed)
On-site Audit2026-09-24 through 2026-09-26 (3 days)
Evidence Review Period2026-06-01 through 2026-09-30 (122 days)
Total Audit Time8 person-days (2 auditors)

2. Audit Methodology

Audit Framework

This audit was conducted per IACS documented procedure IACS-AP-012 "Stage 2 Certification Audit with Cryptographic Evidence" (Rev 1.0), within the scope of CB-4192's ISO/IEC 17021-1 accreditation. Audit planning followed ISO 19011:2018 guidelines.

SWT3 Evidence Verification Method

  1. Portal verification: All 25 anchors checked via sovereign.tenova.io/verify. All returned VALID.
  2. Independent recomputation (5 anchors): One anchor per clause group plus one Annex A control, recomputed using OpenSSL 3.x. All matched.

Temporal Scope

The evidence review period (122 days) covers operational evidence from 2026-06-01 through 2026-09-30. The on-site audit (3 days, 2026-09-24 through 2026-09-26) focused on interviews, observation of operational controls, and verification of document review findings. Of 16 procedures assessed via SWT3, 14 had continuous anchor coverage. Two procedures (AI-DPIA.1 and Clause 9.3 management review) showed gaps documented in the findings below.

Evidence Sufficiency

SWT3 anchors were accepted as objective evidence for operational controls (Clause 8) and performance evaluation (Clause 9.1). Leadership commitment (Clause 5), context analysis (Clause 4), and planning (Clause 6) were assessed through document review and interviews per ISO 19011 methodology.

3. Clause-Level Findings

Clauses 4-5: Context of the Organization and Leadership

Finding: CONFORMING The organization has determined internal and external issues relevant to its AIMS (Clause 4.1), identified interested parties (Clause 4.2), and defined the AIMS scope (Clause 4.3). Top management demonstrates leadership commitment through a signed AI policy (Clause 5.1) with clear accountability assignments (Clause 5.3).

SWT3 Evidence Cited

ProcedureAnchorVerifiedWitness Date
AI-GOV.1 (Governance Attestation) SWT3-E-VULTR-AI-AIGOV1-PASS-1790841600-a2d7f4e1c836 VALID (recomputed) 2026-09-28
AI-GOV.5 (Accountability Assignment) SWT3-E-VULTR-AI-AIGOV5-PASS-1790755200-e8c3b1f7a294 VALID 2026-09-27

Supplementary evidence: AI policy (v2.3, dated 2026-04-01) reviewed. Interview with AIMS Manager (IACS-INT-4192-01) confirmed leadership review of AI risks at quarterly board meetings.

Clause 6: Planning

Finding: CONFORMING Risk assessment methodology is documented and implemented (Clause 6.1). AI system impact assessments are conducted for all high-risk deployments (Clause 6.1.2). AIMS objectives are measurable and tracked (Clause 6.2).

SWT3 Evidence Cited

ProcedureAnchorVerifiedWitness Date
AI-RISK.1 (Risk Identification) SWT3-E-VULTR-AI-AIRISK1-PASS-1790668800-c4f8a2e7b163 VALID (recomputed) 2026-09-26
AI-IMPACT.1 (Impact Assessment) SWT3-E-VULTR-AI-AIIMPACT1-PASS-1790582400-b7d1e5f3a948 VALID 2026-09-25

Supplementary evidence: Risk register reviewed (23 AI-specific risks, 6 classified high-impact). Interview with AIMS Manager confirmed risk assessment conducted before each new model deployment.

Clause 7: Support

Finding: CONFORMING (with observation) Resources are allocated for AIMS operation (Clause 7.1). Personnel involved in AI system development and deployment demonstrate awareness of the AI policy (Clause 7.3). Documentation is controlled (Clause 7.5).
Observation (OBS-001) While clinical staff demonstrate awareness of AI system capabilities and limitations, no formal AI competence framework exists (Clause 7.2). Training records show ad-hoc training but no structured competence assessment for personnel interacting with ClinAssist. We recommend formalizing an AI competence matrix with defined competence levels and assessment criteria.

SWT3 Evidence Cited

ProcedureAnchorVerifiedWitness Date
AI-HITL.1 (Human Oversight) SWT3-E-VULTR-AI-AIHITL1-PASS-1790496000-f1a9c3e7b452 VALID 2026-09-24

Supplementary evidence: Interview with Chief Medical Officer (IACS-INT-4192-02) confirmed clinical staff understand override procedures. Training records reviewed (12 sessions in audit period, no formal competence assessment).

Clause 8: Operation

Finding: CONFORMING AI system lifecycle processes are planned, implemented, and controlled (Clause 8.2). Data management meets quality requirements (Clause 8.4). Third-party AI components are evaluated and documented (Clause 8.5). The organization maintains model integrity verification and supply chain provenance tracking.

SWT3 Evidence Cited

ProcedureAnchorVerifiedWitness Date
AI-MDL.1 (Model Integrity) SWT3-E-VULTR-AI-AIMDL1-PASS-1790409600-d3e7f1a2c894 VALID 2026-09-23
AI-DATA.1 (Data Provenance) SWT3-E-VULTR-AI-AIDATA1-PASS-1790323200-a8c4e9f7b231 VALID (recomputed) 2026-09-22
AI-CHAIN.1 (Supply Chain) SWT3-E-VULTR-AI-AICHAIN1-PASS-1790236800-b2f6c8d1e743 VALID 2026-09-21
AI-SAFE.1 (Safety Classification) SWT3-E-VULTR-AI-AISAFE1-PASS-1790150400-e4a7c3f9b128 VALID 2026-09-20

Supplementary evidence: Interview with AI Development Lead (IACS-INT-4192-04) confirmed model versioning, data pipeline documentation, and supplier evaluation procedures. On-site observation of deployment pipeline confirmed automated integrity checks.

Clause 9: Performance Evaluation

Finding: CONFORMING (partial, with nonconformity) Monitoring, measurement, and analysis are operational (Clause 9.1). Internal audit has been conducted within the required cycle (Clause 9.2). Drift detection and audit log integrity are continuously witnessed.
Minor Nonconformity (NC-001) Management review (Clause 9.3) was not conducted within the required 12-month cycle. The last management review was completed on 2025-08-15, exceeding the 12-month interval by 46 days at audit close. Management review inputs and outputs per Clause 9.3.2 and 9.3.3 cannot be evaluated without a current review. Corrective action: complete management review within 45 days.

SWT3 Evidence Cited

ProcedureAnchorVerifiedWitness Date
AI-DRIFT.1 (Drift Detection) SWT3-E-VULTR-AI-AIDRIFT1-PASS-1790064000-c7e1f3a9b462 VALID 2026-09-19
AI-AUDIT.1 (Audit Log Integrity) SWT3-E-VULTR-AI-AIAUDIT1-PASS-1789977600-a4b8c2f6e137 VALID (recomputed) 2026-09-18

Supplementary evidence: Internal audit report (IA-2026-003, dated 2026-07-12) reviewed. Management review minutes from 2025-08-15 reviewed but overdue for renewal.

Corrective action required: Complete management review per Clause 9.3 covering all required inputs (9.3.2) and producing all required outputs (9.3.3). Submit evidence of completion. Deadline: 45 days (2026-11-22).

Clause 10: Improvement

Finding: CONFORMING Nonconformity management and corrective action processes are established (Clause 10.1). Continual improvement of the AIMS is demonstrated through documented improvement actions (Clause 10.2). Incident reporting and response capabilities are operational.
Observation (OBS-002) While AI-DPIA.1 (Data Protection Impact Assessment) has been conducted for ClinAssist v2.0, the DPIA was completed prior to the audit period (2026-02-10) and has not been refreshed since a significant model update in 2026-07. We recommend updating the DPIA to reflect the current model version per Clause 10.2 continual improvement requirements.

SWT3 Evidence Cited

ProcedureAnchorVerifiedWitness Date
AI-INCIDENT.1 (Incident Reporting) SWT3-E-VULTR-AI-AIINCIDENT1-PASS-1789891200-f8d2a6c4e319 VALID 2026-09-17
AI-IR.1 (Incident Response) SWT3-E-VULTR-AI-AIIR1-PASS-1789804800-b3e7f1c9a452 VALID 2026-09-16
AI-FAIR.1 (Bias Detection) SWT3-E-VULTR-AI-AIFAIR1-PASS-1789718400-d1a4c8f7e236 VALID 2026-09-15
AI-EXPL.1 (Explainability) SWT3-E-VULTR-AI-AIEXPL1-PASS-1789632000-a9c2e6f4b718 VALID 2026-09-14

4. Annex A Control Assessment Summary

Of the 39 Annex A controls in ISO/IEC 42001:2023, 31 were determined applicable per the organization's Statement of Applicability (SoA). 8 controls were excluded with documented justification (controls related to autonomous decision-making, which ClinAssist does not perform -- all outputs are recommendations requiring clinical confirmation).

ThemeApplicableConformingObservationNonconformity
A.2-A.4: Policies and Governance8800
A.5-A.7: Engineering and Development1091 (OBS-002, DPIA)0
A.8-A.9: Operations8701 (NC-001, carried)
A.10: Interested Parties541 (OBS-001, competence)0
Total312821

5. Corrective Actions Register

IDClause / ControlSeverityDescriptionDeadlineResponsible
NC-001 Clause 9.3 MINOR Complete management review per Clause 9.3 with all required inputs (9.3.2) and outputs (9.3.3). Submit evidence of completion to certification body. 2026-11-22 AIMS Manager, Verdant Health Technologies

Observations

IDClause / ControlDescriptionRecommendation
OBS-001 Clause 7.2 / A.10 No formal AI competence framework for clinical staff interacting with ClinAssist. Develop competence matrix with defined levels and assessment criteria before first surveillance audit.
OBS-002 A.5-A.7 / AI-DPIA.1 DPIA not refreshed after significant model update (2026-07). Update DPIA to reflect current model version. Consider triggering DPIA refresh on major version changes.

6. Evidence Verification Log

ProcedureAnchor (Truncated)Verification MethodResultVerified On
AI-GOV.1...a2d7f4e1c836SHA-256 recomputation (OpenSSL 3.x)MATCH2026-10-02
AI-RISK.1...c4f8a2e7b163SHA-256 recomputation (OpenSSL 3.x)MATCH2026-10-02
AI-DATA.1...a8c4e9f7b231SHA-256 recomputation (OpenSSL 3.x)MATCH2026-10-02
AI-AUDIT.1...a4b8c2f6e137SHA-256 recomputation (OpenSSL 3.x)MATCH2026-10-02
AI-EXPL.1...a9c2e6f4b718SHA-256 recomputation (OpenSSL 3.x)MATCH2026-10-02

Remaining 20 anchors were verified via the public verifier at sovereign.tenova.io/verify. All returned VALID.

7. Conclusion and Certification Recommendation

Audit Outcome: CONDITIONAL CERTIFICATION RECOMMENDED

Verdant Health Technologies' AIMS demonstrates substantial conformity with ISO/IEC 42001:2023. One minor nonconformity (NC-001, management review overdue) was identified. Two observations were recorded for continual improvement.

The certification body recommends:

  1. Issue certification conditional on resolution of NC-001 within 45 days
  2. Submission of management review evidence for verification (no re-audit required for minor nonconformity)
  3. First surveillance audit scheduled for October 2027, with priority review of OBS-001 (competence framework) and OBS-002 (DPIA refresh cycle)

SWT3 cryptographic witness evidence provided continuous operational assurance over the 122-day audit period, complementing traditional audit methods (document review, interviews, on-site observation) with tamper-evident, independently verifiable records of control effectiveness.

8. Signature Blocks

Lead Auditor Dr. Yuki Tanaka
International AI Certification Services
Date: _______________
Technical Expert Andreas Keller
International AI Certification Services
Date: _______________
Quality Reviewer Sarah Chen
International AI Certification Services
Date: _______________

Annex A: How to Read SWT3 Anchor References

SWT3 witness anchors cited in this report follow the format:

SWT3-{TIER}-{PROVIDER}-{UCT}-{PROCEDURE}-{VERDICT}-{EPOCH}-{SHA256_12}

Any anchor can be independently verified at sovereign.tenova.io/verify with no account or authentication. The SHA-256 fingerprint formula is published in the SWT3 Protocol Specification.