Third-Party Risk
AI Supply Chain Risk: The AI Already Inside Your Vendors
Your suppliers are shipping AI features whether you asked for them or not. A practical approach to AI supply chain risk and third-party AI governance for regulated organisations.
When organisations think about AI risk, they usually picture their own projects: the chatbot, the forecasting model, the internal copilot. In my experience, the bigger AI supply chain risk is somewhere else — inside the products your suppliers are already selling you.
The quiet arrival of AI
Over the past two years, almost every major SaaS platform has added AI features. Many arrive switched on by default. Some send your data to third-party model providers. Few were covered by the security assessment you ran when you first onboarded the vendor.
That creates three problems:
- Data flows you didn’t approve. Customer, employee or operational data reaching model providers outside your risk appetite.
- Decisions you can’t explain. AI-generated outputs influencing processes without human review.
- Obligations you can’t evidence. Regulators and auditors asking questions your supplier contracts don’t answer.
Why this matters for regulated entities
For critical infrastructure operators, the SOCI Act’s risk management program obligations explicitly include supply chain hazards. For Defence-industry suppliers, security assessments increasingly examine the AI products an entity relies on. “We didn’t know the vendor added that” is not a defensible position.
A practical approach
You don’t need a new program. You need to extend the one you have.
Ask better questions at onboarding
Add a short AI section to your supplier security questionnaire (or start from my free third-party risk questionnaire template, which already includes AI questions):
- Does the product use AI or machine learning? Which features?
- Is customer data used to train or fine-tune models?
- Which third-party model providers are involved, and where is data processed?
- Can AI features be disabled or scoped at tenant level?
- How are AI-related incidents notified?
Re-assess when features change
Vendor release notes are now a risk signal. Build a lightweight trigger: when a critical supplier announces an AI capability, the relationship owner flags it for review.
Contract for it
Make sure agreements address data use for training, sub-processor disclosure, notification of material AI changes and your right to opt out.
Third-party risk management was always about trust, verified. AI just raised the stakes on the “verified” part.
Start with your top ten
Pick your ten most critical suppliers. For each one, find out whether AI features are active, what data they touch, and whether anyone approved them. The answers are usually enlightening — and they give you a defensible, prioritised starting point.
Frequently asked questions
What is AI supply chain risk?
AI supply chain risk is the exposure created when suppliers, SaaS platforms or service providers use AI on your data or in services you depend on — including data sent to third-party model providers, unexplainable automated decisions and obligations you can’t evidence.
Does the SOCI Act cover AI in the supply chain?
The SOCI Act’s risk management program rules require responsible entities to manage supply chain hazards. AI features in critical suppliers are part of that supply chain and should be assessed accordingly — see the Cyber and Infrastructure Security Centre for official guidance.
How do I start assessing third-party AI risk?
Add AI questions to your supplier questionnaire, re-assess when vendors announce AI features, and update contracts to cover training-data use and sub-processors. If you’d like help building that program, explore third-party risk services.