ISO 42001 Annex A 4.2 Resource documentation
What it says
The organization shall identify and document relevant resources required for the activities at given AI system life cycle stages and other AI-related activities relevant for the organization.
Quoted as published. Everything else on this page is written by Control Stack.
In plain English
Identify and document relevant resources, such as data, tools and people, needed for each AI system life cycle stage and for other relevant AI-related work.
Your organisation must identify and document the relevant resources required for both:
- Life cycle activities: the work at given stages of an AI system's life cycle, from design or purchase through operation to retirement.
- Other AI-related activities relevant to your organisation, such as assessing an AI supplier.
Resources are what the work needs, such as system components, data, tools, computing and people, whoever provides them. Your organisation decides which resources and other activities are relevant, based on its risks and context.
Free. Save your status, tasks and notes.
Why it matters
You cannot assess an AI system's risks or its impacts on people without knowing what it depends on. A documented list exposes hidden dependencies, such as a supplier's model or one specialist, and shows early when a required resource is not available.
How to implement
List the AI systems and activities in scope.
Note each AI system's life cycle stages, and add AI work not tied to one system, such as assessing AI suppliers or monitoring AI use. Your organisation decides which other activities and which resources are relevant, based on its risks and context. Record why you left out any stage or activity.
Identify what each stage and activity requires.
Work through components, data, tools, computing and people, using data flow or system architecture diagrams to show dependencies. Record this as the overall list: Annex A 4.3, Annex A 4.4, Annex A 4.5 and Annex A 4.6 require the detail on data, tooling, computing and people, so leave room for it.
Record who provides each resource and whether it is available.
Note whether it comes from your organisation, a customer or a third party. Where a required resource is not available, change the design or the deployment plan, or record the shortfall and decide what to do about it.
For a bought AI service, document what you can see and ask for the rest.
The supplier holds much of the detail on its models and computing, so use its documentation and record what it will not disclose. Still document your side: the data you send, integrations and the people who run and check the service.
Keep it running
Update the record when an AI system moves to a new life cycle stage, when a supplier, data source or key person changes, and when your organisation takes on a new AI-related activity.
Evidence and audit
- Check 1 of 4What an auditor asks for
Ask for
The resource documentation, compared with supplier contracts, cloud bills and software records.
Look at
Whether AI-related resources those sources show, such as a model service or computing capacity, appear in the documentation.
Good evidence
Every AI-related resource those sources show appears in the documentation, or has a recorded reason why the organisation does not treat it as relevant.
- Check 2 of 4What an auditor asks for
Ask for
The resource record for one AI system.
Look at
Whether resources are set out for the activities at each life cycle stage, and whether components, data, tools, computing and people are all considered.
Good evidence
The record shows the resources required for the activities at each stage, and gives a reason for any stage or any of those five kinds of resource it leaves out.
- Check 3 of 4What an auditor asks for
Ask for
A list of AI work not tied to one system, built from the team's own account and from supplier and project records.
Look at
Whether each activity, such as assessing an AI supplier, has its resources documented.
Good evidence
Each AI-related activity the organisation treats as relevant has its resources documented, and any activity found elsewhere but left out has a recorded reason.
- Check 4 of 4What an auditor asks for
Ask for
A sample of entries, checked against the live environment or the supplier's current documentation.
Look at
Whether each resource still exists as described and is available.
Good evidence
Each sampled entry matches what is in use today, and any required resource that is not available is recorded as a shortfall.
Mappings
Closest match in ASD Information Security Manual: ISM-2135, Recording identifier, owner, identities, credentials and access for each AI agent, which partially overlaps with this control.
ASD Information Security Manual: 2 mappings
- ISM-2135Recording identifier, owner, identities, credentials and access for each AI agentPartially overlaps
This control partially overlaps ISM-2135.
Annex A 4.2 requires the organisation to identify and document the resources needed at each AI system life cycle stage (e.g. people, data, tooling, compute). ISM-2135 relates by requiring detailed documentation of AI-agent-specific operational resources and dependencies (identifiers, ownership, identities/credentials, and accessible tools/data repositories) that are part of what must be resourced and documented to run and govern AI agents. ISM-2135 is narrower and security-focused on access and accountability for agents, whereas Annex A 4.2 is broader resource documentation across all AI activities and life cycle stages.
This control supports ISM-2134.
Annex A 4.2 requires identifying and documenting resources required for AI life cycle stages and AI-related activities. ISM-2134 supports this by establishing a maintained and verified AI agent register, which becomes a practical artefact for capturing AI-agent resources (owners, purposes, and operational dependencies) in a controlled way. ISM-2134 focuses on the governance process for maintaining the register, while Annex A 4.2 focuses on completeness of resource identification/documentation across the AI life cycle.
Mappings show how controls relate across frameworks. They do not imply full equivalence or certification.
Get help with this control
ISO 42001 training
Mindset Cyber, a PECB Authorised Titanium Partner, sells official PECB ISO/IEC 42001 courses that map directly to the AI controls in this library.
Turn this control into your AI policy
PolicyMint writes customised ISO 42001 policy documents for your business, audit-ready and exported in your branding (Word, PDF, Excel). Skip the blank page and the boilerplate that does not fit. First document free.
Draft your ISO 42001 documents



