Question 170
Original unofficial Palo Alto Networks NetSec-Architect scenario: A model serving architecture has independent security controls for prompts and attached documents, but neither receives metadata about the user and data classification. Which test result best demonstrates that the design meets its requirement?
Explanation
Correct Answer
Original fictional security architecture situation: A model serving architecture has independent security controls for prompts and attached documents, but neither receives metadata about the user and data classification. Governing principle: Content patterns alone cannot express user-specific entitlement or business-purpose decisions. The most defensible design response is: Pass approved identity and data classification context into the AI security decision path while protecting that metadata. Validation: Policies make distinct decisions for an authorized analyst and an unauthorized guest submitting the same sample. Sustained assurance: Validate that identity context is fresh, integrity protected, and available during control failure
Incorrect Answers
The proposed design “A test request to an overbroad document index does not produce the planted synthetic secret.” addresses this related architecture topic: select ai security reference architectures, form factors, and data controls. The actual security gap is that a model serving architecture has independent security controls for prompts and attached documents, but neither receives metadata about the user and data classification. Content patterns alone cannot express user-specific entitlement or business-purpose decisions.
“A synthetic prompt trying to reveal protected data is blocked or safely transformed and recorded.” concerns the separate design topic: differentiate prisma airs and ai access capabilities. It would not fix the described weakness: a model serving architecture has independent security controls for prompts and attached documents, but neither receives metadata about the user and data classification. Content patterns alone cannot express user-specific entitlement or business-purpose decisions.
This alternative is linked to the architecture category: classify ai applications, sanction usage, and align security with frameworks. The actual problem is that a model serving architecture has independent security controls for prompts and attached documents, but neither receives metadata about the user and data classification; the controlling distinction is: content patterns alone cannot express user-specific entitlement or business-purpose decisions.