Free template

The 25 AI security questions enterprise buyers ask

Collected from the AI sections of SIG 2026, CSA AI-CAIQ, HECVAT 4 and the custom appendices security teams attach to vendor reviews. Use it to prepare before the questionnaire arrives, or connect your AI accounts and let us answer the verifiable ones.

Describe the AI/ML features in your product and their purpose.your statement
Which third-party AI providers do you use?partly verifiable
Which models are in use, and how are new models approved?partly verifiable
Do you route AI traffic through a gateway or proxy with central controls?partly verifiable
How do you restrict which models your systems may call?partly verifiable
Is customer data used to train or fine-tune any model, by you or your providers?your statement
What contractual terms (DPA, zero data retention) are in place with AI providers?your statement
What is the retention period for prompts and outputs at each provider and in your systems?partly verifiable
Do you scan prompts or outputs for sensitive data (DLP)?partly verifiable
Where is AI data processed and stored (regions)?your statement
Who administers your AI provider accounts, and how is access reviewed?partly verifiable
How are AI provider API keys issued, owned, reviewed and revoked?partly verifiable
How often are AI provider credentials rotated?partly verifiable
How are production AI credentials segregated from development and from individuals?partly verifiable
Is there a human in the loop for AI outputs that affect customers?your statement
How do you prevent and detect prompt injection?your statement
What content safety controls or guardrails are applied to prompts and outputs?partly verifiable
How are AI-related incidents detected, handled and communicated?your statement
List all subprocessors involved in AI features.your statement
Is administrative activity on AI provider accounts logged and reviewed?partly verifiable
Do you maintain an inventory (AI bill of materials) of models and providers?partly verifiable
What cost and usage limits protect against abuse or runaway consumption?partly verifiable
What rate limits apply to AI endpoints?partly verifiable
Which AI governance frameworks do you align to (ISO 42001, NIST AI RMF)?your statement
Can customers opt out of AI features?your statement

Copy as Markdown

# AI security appendix

1. **GAI-01** Describe the AI/ML features in your product and their purpose.
2. **GAI-02** Which third-party AI providers do you use?
3. **GAI-03** Which models are in use, and how are new models approved?
4. **GAI-04** Do you route AI traffic through a gateway or proxy with central controls?
5. **GAI-05** How do you restrict which models your systems may call?
6. **GAI-06** Is customer data used to train or fine-tune any model, by you or your providers?
7. **GAI-07** What contractual terms (DPA, zero data retention) are in place with AI providers?
8. **GAI-08** What is the retention period for prompts and outputs at each provider and in your systems?
9. **GAI-09** Do you scan prompts or outputs for sensitive data (DLP)?
10. **GAI-10** Where is AI data processed and stored (regions)?
11. **GAI-11** Who administers your AI provider accounts, and how is access reviewed?
12. **GAI-12** How are AI provider API keys issued, owned, reviewed and revoked?
13. **GAI-13** How often are AI provider credentials rotated?
14. **GAI-14** How are production AI credentials segregated from development and from individuals?
15. **GAI-15** Is there a human in the loop for AI outputs that affect customers?
16. **GAI-16** How do you prevent and detect prompt injection?
17. **GAI-17** What content safety controls or guardrails are applied to prompts and outputs?
18. **GAI-18** How are AI-related incidents detected, handled and communicated?
19. **GAI-19** List all subprocessors involved in AI features.
20. **GAI-20** Is administrative activity on AI provider accounts logged and reviewed?
21. **GAI-21** Do you maintain an inventory (AI bill of materials) of models and providers?
22. **GAI-22** What cost and usage limits protect against abuse or runaway consumption?
23. **GAI-23** What rate limits apply to AI endpoints?
24. **GAI-24** Which AI governance frameworks do you align to (ISO 42001, NIST AI RMF)?
25. **GAI-25** Can customers opt out of AI features?

See the sample verified page at /t/sample and the rules at /checks.