Risk · R-088 · Governance & Accountability
Incomplete AI inventory
The organisation cannot list the AI systems and models it builds, buys or uses, so risk assessment, oversight and regulatory duties are applied to only part of what it has.
- Severity
- High
- Audience
- Board · Heads-of · Practitioner
- Frameworks
- Principle 5 · Annex A.4 · GOVERN 1 · Principle 1
- Responsible AI domain
- Lifecycle stage
- Reviewed
- 2026-11-10
What it is
An inventory is the record of every AI system in use, with its owner, purpose, risk tier, data, supplier and status. It is the foundation of governance: tiering, validation, monitoring and regulatory classification all start from it. The risk is that it is missing, partial or out of date. AI embedded in bought products, models reused by other teams and tools adopted by individuals are the usual omissions. Use that has not been approved at all is covered in shadow AI use.
How it happens
- The definition of what to record is narrow or unclear.
- Registration depends on teams volunteering.
- Suppliers add AI features to products already in use.
- Entries are created at launch and never updated.
Why it matters
Governance reaches only what is listed. An organisation that cannot enumerate its AI cannot say which systems are high-risk, which process personal data, or which depend on a given supplier, and cannot answer a regulator who asks.
Warning signs
- The inventory's count has not changed in months while adoption has grown.
- Procurement and expense records show AI products that are not listed.
- Entries lack an owner, a tier or a review date.
- Different functions hold different lists.
Who owns it
A named governance function owns the inventory and its completeness. Each system owner is accountable for registering and updating their entries. Internal audit tests completeness.
In practice
A regulator asks an insurer which of its systems would be high-risk under the EU AI Act. The inventory lists eleven models built in-house. A review of supplier contracts finds twenty-three products with AI features, including the one used to screen job applicants. The same pattern generalises to any organisation whose record of AI was built from what teams chose to report.
Mitigating controls
Primary controls
- AI inventory and registration Detective
- Records every system with owner, purpose, tier and status and reconciles the record. All AI systems and models, whether built, bought or embedded in other products, are recorded with owner, purpose, tier, data, supplier and status, and the record is reconciled periodically.
- Shadow AI discovery Detective
- Finds systems in use that the record lacks. Network, procurement, expense and software records are reviewed on a set cycle to find AI tools and features in use that are absent from the inventory.
Supporting controls
- AI use-case intake and approval Preventive
- Every approved use enters the record at source. Every proposed AI use is registered, screened, tiered and approved by a named authority before build or purchase begins, and again before it goes live.
- Embedded AI feature review Detective
- Finds AI switched on inside existing products. Supplier release notes, terms and product settings are monitored for new AI features in products already in use, and each feature is assessed and approved before it is enabled.
Framework and clause references
| Framework | Clause | Title |
|---|---|---|
| Code of Practice for the Cyber Security of AI | Principle 5 | Identify, track and protect your assets Asset identification. |
| ISO/IEC 42001:2023 | Annex A.4 | Resources for AI systems Documented resources for AI systems. |
| NIST AI Risk Management Framework (AI RMF 1.0) | GOVERN 1 | Policies, processes and practices for AI risk are in place and effective Inventory is named as a governance practice. |
| SS1/23: Model Risk Management Principles for Banks | Principle 1 | Model identification and model risk classification A complete model inventory. |
Sources
- SS1/23: Model Risk Management Principles for Banks Bank of England PRA · official text
- ISO/IEC 42001:2023 – AI management systems ISO/IEC · framework
- NIST AI Risk Management Framework (AI RMF 1.0) NIST · framework
Related terms
How to cite
Axiom Verity (2026). Incomplete AI inventory. Reference Library, R-088. https://axiom-verity.com/risks/incomplete-ai-inventory. Reviewed 2026-11-10.
Found an error or a better source? Report a correction.