ISO 42001 Annex A 4.4 Document the tooling resources used for AI systems
What it says
As part of resource identification, the organization shall document information about the tooling resources utilized for the AI system.
Quoted as published. Everything else on this page is written by Control Stack.
What it means
In plain EnglishYour organisation must keep written records of the tools used to build, run and support each AI (artificial intelligence) system, as part of identifying the resources your AI management system (AIMS) depends on.
AI (artificial intelligence) systems are built and run using a range of tools: software libraries, development platforms, machine learning frameworks, data labelling tools, model training environments, testing tools and the third-party services that support them. This control asks your organisation to write down which of these tools each AI system actually uses. The reason is simple. If you do not know what your AI is built on, you cannot keep it secure, maintain it, fix it when it breaks, or replace a tool that becomes unsupported or risky. Documenting tooling resources is part of a wider step in the ISO 42001 AI management system (AIMS) called resource identification, where you record all the people, data, computing power and tools your AI depends on. For tools specifically, you keep an up to date list or register that names each tool, what it is used for, who provides it, and which AI system relies on it. This does not need to be complicated. A maintained spreadsheet or register is often enough. The point is that someone can look at any AI system in your organisation and quickly see the tools behind it, so the system can be managed responsibly over its whole life.
Free. Save your status, tasks and notes.
Why it matters
Without documented tooling, the organisation cannot maintain, secure or replace the tools its AI depends on, and one unsupported tool can break or expose several systems unnoticed.
How to implement
The AI or technical lead should build a tooling register that lists every tool used to develop, train, test, deploy and run each AI system, recording the tool name, its purpose, the provider or vendor, the version where relevant, and which AI system uses it.
Development and data science teams should record the tooling they use at the point they adopt it, so new libraries, frameworks, platforms or labelling tools are captured as the AI is being built rather than reconstructed from memory later.
The person who owns the AI management system should set a simple rule that the tooling register is updated whenever a tool is added, upgraded or retired, and should review it at planned intervals to confirm it still matches what is actually in use.
Procurement and IT should feed any AI related software, cloud services or platform subscriptions they purchase into the tooling register, so externally sourced tools are documented alongside ones built or chosen in house.
The AI lead should link each tool in the register to the AI system it supports, so that if a tool is found to be insecure, unsupported or non-compliant, the organisation can immediately see which AI systems are affected.
Keep it running
Keep the tooling register current: update it whenever a tool is added, upgraded or retired, and review it at planned intervals so it always matches the tools in real use.
Evidence and audit
- Check 1 of 2What an auditor asks for
Ask for
the documented tooling register or resource records for AI systems, and check that documenting tools is treated as a defined part of resource identification rather than something done ad hoc.
Look at
whether the records name the specific tools used to develop, train, test, deploy and run each AI system, including the provider and what each tool is used for, not just a vague mention that tools exist.
- Check 2 of 2What an auditor asks for
Ask for
to trace one or two live AI systems back to their documented tooling, and check the listed tools genuinely match the development platforms, frameworks and services those systems actually use.
Look at
the update history or version dates on the register to confirm it is kept current when tools are added, upgraded or removed, rather than written once and left to go stale.
Good evidence
a maintained, owned tooling register where every AI system can be linked to its tools, recently reviewed, with clear accountability for keeping it accurate.
Mappings
Closest match in ISO/IEC 27001: A.5.9, Inventory management of information and associated assets, which partially overlaps with this control.
ISO/IEC 27001: 1 mapping
This control partially overlaps A.5.9.
Annex A 4.4 requires the organisation to document the tooling resources utilised for each AI system as part of AIMS resource identification. Annex A 5.9 requires an inventory of information and other associated assets (including owners), which can encompass AI tooling as an asset class (e.g. software tools, platforms, and services) and provides governance structure for maintaining that inventory. Annex A 5.9 is broader and not AI-specific, whereas Annex A 4.4 is narrower but requires AI-system linkage (which tools support which AI system and for what lifecycle activities).
ASD Information Security Manual: 3 mappings
This control partially overlaps ISM-1643.
Annex A 4.4 requires documented information about tooling resources used by an AI system to support resource identification within the AIMS. ISM-1643 strengthens this by requiring software registers to include versions and patch histories, which are important attributes of AI tooling (e.g. framework versions, inference servers, data processing tools) and support traceability and reproducibility. ISM-1643 is focused on patch/version recordkeeping for general software, while Annex A 4.4 is focused on the AI system's tooling set (including non-installed or external tools such as managed services, MLOps platforms, and evaluation tooling).
This control partially overlaps ISM-2134.
Annex A 4.4 requires documenting information about the tooling resources used for an AI system as part of AIMS resource identification. ISM-2134 overlaps by requiring an AI agent register, which often includes operational dependencies and tooling. ISM-2134 focuses on AI agents, while Annex A 4.4 covers all tooling resources for AI systems, not limited to agents.
- ISM-2135Recording identifier, owner, identities, credentials and access for each AI agentPartially overlaps
This control partially overlaps ISM-2135.
ISM-2135 requires an AI agent register entry listing the user accounts/credentials used by the agent and the tools it can access. Annex A 4.4 involves documenting tooling resources for AI systems, which helps organisations systematically identify tools for recording and governance. While Annex A 4.4 focuses on inventorying tooling resources generally, ISM-2135 applies that information to agent-level identity and access traceability.
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 runs PECB-accredited ISO/IEC 42001 training that maps 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