SC-AI-003 · Governance

AI system inventory

Medium effortCovered by ISO 27001: StrongSatisfies 21 obligations in 12 policies

What it means

  • A single, up-to-date list of every AI system the company builds, buys or uses, including AI features switched on inside SaaS tools.
  • Each entry records: purpose, owner, users affected, model and vendor, data used, dependencies, risk tier, where it runs, go-live date, and open issues or grievances.
  • For platforms, it also flags features that create or spread synthetic content (deepfakes, AI images, voice clones), since these carry their own legal duties.

Why it matters

  • You cannot govern what you cannot see. Every other control (risk tiering, testing, monitoring, audit) starts from this list.
  • RBI FREE-AI expects an inventory updated at least half-yearly and ready for supervisors.
  • The 2026 IT Rules amendment requires platforms to know which features generate synthetic content.

Checklist

  1. Define what counts as an AI system for the inventory (link to the definition in SC-AI-001 AI governance policy approved by the board or top management).
  2. Run a discovery sweep: survey business heads, check procurement and expense records for AI tools, scan SaaS admin consoles for AI features, check cloud accounts for model endpoints.
  3. Create the register (GRC tool, CMDB extension or controlled spreadsheet) with mandatory fields listed above.
  4. Assign each entry a risk tier via SC-AI-004 AI risk classification and use-case tiering.
  5. Flag synthetic content features separately.
  6. Make intake mandatory: no new AI use goes live until it is in the register.
  7. Reconcile at least every six months and after any major change; record the review.

Evidence examples

  • Exported AI inventory with owner, tier and last-reviewed date for each entry.
  • Half-yearly reconciliation sign-off by the AI risk owner.
  • Procurement or intake form that requires an inventory ID before purchase approval.
  • Discovery evidence, such as a SaaS admin report showing AI features enabled or disabled.

Common pitfalls

  • Listing only in-house models and missing vendor AI and embedded SaaS features.
  • A one-time exercise that is never refreshed.
  • Missing owner or risk tier fields, so the list cannot drive other controls.
  • Keeping the AI list separate from the IT asset register with no link between them.

Satisfies

Standards crosswalk

Clause references only. Check the standard itself for the exact wording.

ISO/IEC 42001

  • 4.3 Scope of the AI management system
  • A.4.2 Resource documentation
  • A.4.3 Data resources
  • A.4.4 Tooling resources
  • A.4.5 System and computing resources

ISO/IEC 27001

  • A.5.9 Inventory of information and other associated assets
  • A.5.12 Classification of information

SOC 2

  • CC6.1

Key takeaways

  • The inventory is the foundation; build it before anything else.
  • Strong reuse from ISO 27001: extend your existing asset inventory (A.5.9) with AI-specific fields rather than starting a new register.
  • Include shadow and embedded AI, and refresh at least half-yearly.

Find your gaps

Our free compliance gap assessment shows which of these controls you already run.

Start the gap assessment

Have us build it with you

Our AI Governance service implements controls like this one and produces the evidence auditors expect.

Explore AI Governance