AI Control Library / CF-03: AI Supply Chain & Third-Party Risk
Designation: Core | Control types represented: Operational
Control requirement
Review AI supply chain agreements at an approved interval and at renewal or material change to confirm that accepted terms, service behavior, responsibilities and intended use remain aligned.
Why this matters
An agreement suitable for an internal assistant may be inadequate after external action-taking is introduced. Renewals and provider policy updates can also alter data or recovery assumptions without a new procurement project.
How the control types work together
Operational controls establish the supplier assessment, ownership, agreement and review processes described here. They work when the assigned parties perform the required activities and use evidence to guide decisions.
When it applies
Active AI contracts, service schedules, licenses and standard terms supporting consequential or sensitive organizational uses.
Tailor review depth to information sensitivity, granted authority and business dependence. Record what the organization can test directly, what relies on provider evidence and how material visibility gaps affect the acceptance decision.
Ownership
The supplier relationship owner coordinates review with legal, procurement, system owners and relevant control specialists.
Implementation guidance
- Maintain authoritative agreements, versions, renewal dates and change-notification channels linked to the service inventory.
- Compare current terms with actual data flows, permissions, use cases and shared responsibilities; identify material mismatches.
- Review evidence rights, incident support, service continuity and exit practicality, considering findings and supplier performance.
- Record renewal, amendment, restriction or termination decisions, update operating conditions and verify implementation of resulting changes.
How to verify it
Use representative deployment records and authorized controlled tests. Establish the assessed service, artifact version and acceptance conditions before interpreting results; do not extend evidence beyond the boundary it covers.
| Test | Expected result |
|---|---|
| Inspect a material term update | It receives an applicability and impact review before continued affected use. |
| Compare the agreement with live integrations | Permissions and information use stay within accepted conditions. |
| Trace a renewal decision | Open findings and exit limitations influence approval or treatment. |
Record a finding when a required result cannot be demonstrated or actual practice differs materially from the accepted conditions. Assign corrective action and verify the affected boundary before closing the issue.
Evidence to retain
- Agreement and renewal register
- Version and change notices
- Review findings
- Renewal, amendment and implementation decisions
Protect supplier-confidential information, credentials and personal data in these records. Preserve enough version and scope information to establish which service or artifact the evidence supports.
If the control fails
Escalate unacceptable changes or unresolved agreement gaps to the relationship and business owners. Restrict affected processing or prepare an evaluated exit rather than silently carrying forward obsolete approval.
Example in practice
Hypothetical example: At renewal, a service introduces an additional processing region. The review checks the organization’s data requirements and either establishes acceptable terms or removes the affected datasets from the service.
Related controls
- CF03-007: AI Supply Chain Contractual Requirements
- CF03-009: AI Supply Chain Compliance Assessment
- CF01-S02: Change Agreements for AI Systems
Framework mappings
- CSA AICM (STA-11)
- NIST AI RMF (MANAGE 2.2)
- EU AI Act (Art. 28)
- ISO 42001 (Cl. 9.2, A9.2)
Review and maintenance
Review at the interval approved by the control owner and after relevant supplier, component, contract, access or deployment changes. Reassess material incidents and new evidence, updating acceptance conditions and verifying corrective actions where required.
Back to AI Supply Chain & Third-Party Risk
Version 1.0