フリー問題

問題文

For an internally built AI application in Azure, which mapping of concerns to controls is correct?

選択肢

1. The Defender plan for AI services for attack detection and for content risk, because the same signals reveal both malicious activity and unsafe responses
2. Guardrails in Foundry for content risk and for request rationing, because every request and response passes through them and can be counted and shaped there as well, so that a single control point covers both the safety of the content and the volume of traffic reaching the model
3. Request rationing and routing with the AI gateway in API Management, content risk with guardrails in Foundry, attack detection with the Defender plan for AI services, and data risk with the Purview posture solution
4. The AI gateway in API Management for request rationing and routing, and for content risk as well, because all traffic passes through it and can be inspected there