ISO 42001
AI Management System
The assessment does not stop at checking whether a policy or document exists. Controls are assessed across five stages — from design intent through to demonstrated effectiveness — so the outcome distinguishes “we have a policy” from “we have an operating control that can be demonstrated”.
Methodology
- 01
Identify AI
Establish where AI is used across the organisation.
- 02
Classify Use Cases
Classify the identified AI use cases.
- 03
Assess Risk
Assess the risk attaching to each classified use case.
- 04
Define Controls
Define the controls the assessed risk calls for.
- 05
Implement Governance
Put the governance for those controls in place.
- 06
Validate
Confirm the controls operate as defined.
- 07
Monitor
Keep the management system under review.
Approach to testing
- Requirement — what does the applicable standard, regulation or framework require?
- Control — what control has the organisation established?
- Implementation — how is the control actually implemented?
- Evidence — what evidence demonstrates that the control operates?
- Risk — what happens if the control is ineffective or absent?
- Action — what needs to be changed?
- Validation — has the corrective action actually addressed the issue?
Types of assessment
Black-Box
Assessment begins with limited organisational information, to provide an independent perspective of the governance environment.
Grey-Box
Selected organisational documentation, process information and evidence are provided for structured assessment.
White-Box
Full documentation, evidence, stakeholder and process access is provided for detailed control validation.
Hybrid
Combines independent assessment techniques with detailed evidence and stakeholder validation.
Frameworks and standards
- ISO/IEC 42001
- The AI management system standard the governance is built to.
- AI Governance Control Framework
- The control framework applied against it.
- AI Risk Assessment Toolkit
- Applied to each classified use case.
- AI System / Use-Case Inventory
- The inventory the assessment runs from.
- AI Governance Policy Toolkit
- Policy set for the management system.
Tools used
Tooling is where testing starts, not where it ends. Every automated result is reproduced by hand before it reaches a report.
AI Governance Toolkit
AI inventory, AI risk assessment, AI governance, AI policy, AI vendor assessment and AI control assessment.
AI System / Use-Case Inventory
Establishes what is in scope before any control is chosen.
PlyoGRC
Where appropriate, the toolkits are supported through PlyoGRC for control, evidence, risk and compliance management.
Checklist approach
The checklist is the floor, not the ceiling. It guarantees coverage so nothing standard is missed; the findings that matter usually come from what a tester does after it is complete.
Foundation
- AI governance framework
- AI system inventory
- AI use-case classification
Risk and policy
- AI risk assessment
- AI policies
- AI roles and responsibilities
Lifecycle and third parties
- AI lifecycle governance
- AI vendor governance
- Human oversight
Operation and assurance
- AI incident governance
- AI control implementation
- AI management review
- AIMS readiness
How CSS tests
A unified swarm of agents, for blind spot detection
AI agents drive several testing tracks against the same target at once, then cross-check each other. A single tester works one hypothesis at a time; parallel agents cover the space a sequential pass leaves behind.
Framework Mapping Agent — maps requirements and controls across applicable frameworks.
Policy Analysis Agent — identifies potential missing, inconsistent or outdated requirements.
Evidence Analysis Agent — associates evidence with applicable controls and identifies evidence gaps.
Risk Analysis Agent — identifies recurring risk themes and potential control weaknesses.
Blind-Spot Detection Agent — looks for issues that may not be immediately visible through conventional checklist assessment.
Executive Reporting Agent — helps transform detailed assessment information into concise management reporting.
AI-assisted analysis supports the assessment team but does not replace professional judgement. Material findings, risk conclusions and recommendations are reviewed and validated by CyberSmithSECURE professionals.
Why this differs
What CSS does that most vendors do not
Every one of these is checkable. Ask any vendor for the same and compare the answers.
Beyond the checklist
Structured checklists and framework mappings establish coverage, but the assessment continues through implementation, evidence, risk, action and validation.
Operating control, not documentation only
Evidence is validated across five stages: Designed — is it appropriately designed? Implemented — has it been implemented? Operating — is it actually performed? Evidenced — can operation be shown? Effective — is it achieving its goal?
Depth selected per engagement
Black-box, grey-box, white-box or hybrid, chosen on the purpose, scope and risk of the engagement rather than applied uniformly.
Human-in-the-loop AI assistance
AI-assisted analysis supports the assessment team but does not replace professional judgement. Material findings, risk conclusions and recommendations are reviewed and validated by CyberSmithSECURE professionals.
What you receive
A working management system, not a folder of documents
The target state is that owners know what they must do, management knows what decisions are pending, and evidence exists to demonstrate that controls operate. Outputs are grouped by who uses them.
Executive layer
Scope, risk posture, roadmap, management decisions, KPI/KRI, readiness summary
GRC layer
Risk register, Statement of Applicability, policies, procedures, ownership, evidence map, action tracker
Assurance layer
Internal audit, findings, CAPA, management review, certification-readiness assessment
Operational layer
Control records, recurring reviews, awareness, supplier / access / incident / continuity evidence as applicable
Governance cadence established
- Monthly
- Risk / action review, evidence status, control exceptions, material incidents
- Quarterly
- Risk trend, supplier / control reviews, KPI/KRI, management action tracking
- Annual
- Internal audit programme, management review, ISMS objectives, risk refresh, improvement plan
For this engagement specifically
- Control owners
- IT teams
- Security teams
- Compliance teams
- Process owners
- Auditors
- Key risks
- Significant gaps
- Business impact
- Priority actions
- Ownership
- Target timelines
Next
Scope this assessment
Most scopes are settled in one call. Tell us what the application does and who uses it, and we will tell you what testing it properly involves.