AI Governance and Compliance
preparation for ISO/IEC 42001 + AI Act, PECB IMS2 Method
Methodology – PECB IMS2
(Integrated Implementation Methodology for Management Systems and Standards)
Objective
Implement an AI Management System (AIMS) compliant with ISO/IEC 42001, produce auditable evidence, and develop a roadmap for AI Act compliance (classification + requirements + compliance dossier) in order to be ready for the certification audit.
1) Expected results (deliverable commitment)
At the end of the assignment, you will have :
- An operational AIMS (governance, roles, processes, metrics) covering the defined scope.
- A registry of AI systems and a mapping of use cases.
- An AI risk register including risk treatments, acceptance criteria, and controls.
- The AI policy and documentation (procedures) required by ISO 42001.
- An evidence pack to demonstrate compliance during an audit.
- An internal audit and a management review conducted (with reports).
- A certification readiness assessment (pre-audit) and a corrective action plan.
- An AI Act roadmap : system classification, mapping of obligations, list of evidence to be compiled, and compliance initiatives.
2) Scope (what the offer covers)
2.1 Scope of “AIMS ISO/IEC 42001”
- Governance : leadership, responsibilities, committees, objectives, and KPIs.
- AI risk management : identification, analysis, mitigation, and monitoring.
- AI lifecycle : design/selection, training/evaluation (if applicable), deployment, monitoring, change management, decommissioning.
- AI supplier/third-party management : requirements, evaluation, contract terms, monitoring.
- Documented information management : policies, procedures, records. Competencies & awareness.
- Monitoring, internal audit, management review, continuous improvement.
2.2 Scope of the “AI Act” (compliance path)
- Classification of AI systems (by use case / product / module).
- Mapping of applicable requirements (transparency, documentation, governance, human oversight, robustness/cyber, data, monitoring, etc.).
- Compliance file : structure, responsibilities, expected evidence, remediation plan.
This offering is designed to provide structured preparation for the AI Act. Full compliance depends on the type of system (e.g., “high-risk,” GPAI), the context, and technical choices. The engagement provides the roadmap, the compliance dossier, and the compliance plan.
3) Implementation process (PECB IMS2 aligned with ISO/IEC 42001)
Phase 1 — Define & Establish
Objective : Provide a framework, secure buy-in, define the scope and documentation requirements.
Activities :
1. Leadership & Approval : Sponsor, project governance, objectives, milestones.
2. Roles & Responsibilities : RACI (SMIA Owner, AI Managers, Risk, DPO, Security, Operations).
3. Context & Stakeholders : customer expectations, regulators, suppliers, risks.
4. SMIA Scope : entities, products, use cases, data, environments.
5. Analysis of Current State : AI practices, SDLC/ML lifecycle, security, privacy, GRC.
6. AI Policy : principles, commitments, acceptance criteria, escalation.
7. AI Risk Management : methodology, matrices, thresholds, workflow, registers.
8. Statement of Applicability (SoA 42001) : selected controls/justification.
Phase 1 Deliverables
- Project charter + governance + RACI
- SMIA scope
Initial AI systems register (v1) - AI policy (v1)
- AI risk methodology + risk register (v1)
- 42001 Statement of Applicability (v1)
- Project plan + evidence plan (structure)
Phase 2 — Implement and Operate
Objective : Implement measures, develop procedures, and launch operations.
Activities :
1. Selection & design of measures : AIMS controls, supplier requirements, validation criteria, monitoring, logs.
2. Implementation : Procedures + basic tools (templates, workflows, tickets).
3. Management of documented information : document repository, versions, evidence.
4. Communication : AI usage rules, committees, decision-making processes.
5. Skills & awareness : targeted training (executives, product, data/AI).
6. AI operations management : operations, monitoring, AI incidents, changes.
Phase 2 Deliverables
- AI Registry (v2) + use case sheets
- Key procedures : AI lifecycle, validation/release, monitoring, AI incidents, changes, decommissioning, AI vendor management, data management for AI (governance)
- Evidence catalog + record templates (logs, reviews, validations)
- SMIA KPI/OKR matrix + dashboard (structure)
- Training plan + materials + certificates
Phase 3 — Monitor and Review
Objective : To demonstrate that the system is operational and being effectively managed.
Activities :
1. Monitoring, measurement, analysis, and evaluation : KPIs, controls, periodic reviews.
2. Internal audit (ISO/IEC 42001) : plan, checklists, interviews, findings.
3. Management review : decisions, trade-offs, improvement plan, resources.
Phase 3 Deliverables
- Internal audit plan + internal audit report
- Management review minutes + decisions + action plan
- Update of SoA, AI risks, logs, evidence
Phase 4 — Maintain and Improve
Objective : Correct, stabilize, and prepare for the certification process.
Activities :
1. Non-conformity handling : corrective actions, effectiveness verification.
2. Continuous improvement : optimization of controls, maturity, automation.
Phase 4 Deliverables
- Non-conformity log + corrective actions + closure evidence
- “Certification Readiness” package : final evidence file + audit transition plan + checklist
4) AI Act component (integrated as work progresses)
4.1 AI Act Workshops
- Inventory of relevant AI systems (aligned with the AI registry).
- Classification (relevant categories based on use cases).
- Mapping of obligations (by system) and responsibilities.
- Evidence plan (documentation, traceability, transparency, oversight, robustness, etc.).
4.2 AI Act Deliverables
- Classification Matrix & Requirements
- Compliance Roadmap (quick wins / structural initiatives)
- Structure of the compliance dossier + list of supporting documentation to be compiled
- Contractual recommendations (AI providers / subcontractors) as needed
5) Project organization
Workshops (typical)
- Kick-off + Scope Definition
- AI Registry & AI Risk Workshops
- AI Lifecycle & Operations Workshops
- AI Suppliers/Third-Party Workshop
- AI Act Workshop: Classification & Requirements
- Internal Audit + Management Review + Readiness
Client-side roles (minimum)
- Sponsor (Management)
- AIMS Lead (Owner)
- Product/AI Lead (CTO/Head of AI or equivalent)
- Risk/Compliance Lead and/or DPO
- Security Lead (CSO) if available
6) Duration (to be adjusted based on the scope)
- Standard : 8 to 12 weeks for a limited “product/platform” scope and 3–10 use cases.
- Extended : 12 to 16 weeks for multi-team, multi-product projects or those involving numerous or critical use cases.
7) What drives value (certification + compliance)
- Accelerated sales to key accounts : fewer back-and-forth exchanges regarding “trust and assurance,” and improved responses to RFPs.
- Risk reduction : structured AI risks, traceable decisions, and implemented controls.
- Single, auditable framework : centralized evidence, internal audit, and management review.
- AI Act-ready : classification + requirements + compliance plan and evidence to be provided.
8) Options
- “AI Vendors” Package : due diligence + contract clauses + third-party monitoring
- “AI Architecture & Security” Package : security review, hardening, logging, operational requirements
- “Pre-Penetration Test” Package : certification audit simulation
- “Scale” Package : extension of the AIMS to other products/use cases
