AI Control Library / CF-03: AI Supply Chain & Third-Party Risk
Designation: Core | Control types represented: Governance and Operational
Control requirement
Include risk-based AI supply chain requirements in the applicable agreements, addressing permitted use, data handling, security responsibilities, changes, incidents, evidence, service resilience and exit. Identify requirements the supplier will not accept.
Why this matters
A questionnaire response does not establish an enforceable service commitment. Important assumptions need to be reflected in the agreement or addressed through an explicit limitation on how the service may be used.
How the control types work together
Governance controls establish applicable binding external rules and requirements. Operational controls translate those requirements into supplier selection, agreements, evidence and continued-use decisions.
When it applies
AI services and components obtained under negotiated contracts, standard terms or licenses; do not assume free or open distribution removes use restrictions.
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
Procurement and the service owner establish requirements with legal, security and privacy reviewers; the authorized business owner accepts residual commercial risk.
Implementation guidance
- Derive required clauses and acceptance conditions from the intended use, applicable external obligations and supplier risk assessment.
- Address prompt and output use, retention, training use, subprocessors, access, incident notification, change handling and evidence rights as relevant.
- Define service and exit responsibilities, including data return or deletion, continuity limits and handling of outstanding work.
- Compare final terms with requirements, record deviations and impose operational restrictions when negotiation cannot secure a necessary protection.
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 |
|---|---|
| Trace a material risk requirement | An applicable commitment or explicitly approved restriction addresses it. |
| Compare marketing statements with signed or accepted terms | Approval relies on the actual agreement. |
| Review termination arrangements | Data, access and continuity responsibilities are actionable. |
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
- Contract requirement schedule
- Reviewed terms and licenses
- Deviation and risk decisions
- Service and exit commitments
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 conflicting terms or unmet mandatory conditions to legal and the accountable owner. Withhold sensitive access or the affected use until an acceptable arrangement is established.
Example in practice
Hypothetical example: A supplier’s product page promises enterprise privacy, but default terms allow additional data use. Legal and procurement resolve the applicable service schedule before the buyer connects its internal document store.
Related controls
- CF03-006: AI Supply Chain Risk Assessment
- CF03-008: AI Supply Chain Agreement Review
- CF03-010: AI Supply Chain Provider Compliance Requirements
Framework mappings
- CSA AICM (STA-10)
- NIST AI RMF (GOVERN 6.1)
- EU AI Act (Art. 28, Art. 30)
- ISO 42001 (Cl. 8.3, 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