Skip to main content
  1. Public practice questions
  2. AB-650 Practice Exams | Microsoft 365 and AI Services Administrator Associate
  3. AB-650 Practice Exam 5

Question 31 — AB-650 Practice Exam 5

A Microsoft Purview DLP policy is assigned to members of the Finance group, detects credit card numbers, and monitors endpoint file-upload activity. An AI agent associated with a different group sends a credit card number directly to a human reviewer. The Finance endpoint policy does not generate the alert. What explains this result? (Select one answer)

  1. The policy cannot detect credit card numbers when they appear in AI-generated messages.

  2. The policy evaluates the message but suppresses the alert because an AI agent sent it.

  3. The activity and assigned scope do not match the policy, even though the sensitive information type matches.

  4. The policy requires the credit card number to appear in both an uploaded file and an agent message.

Correct answers

  1. The activity and assigned scope do not match the policy, even though the sensitive information type matches.

Explanations

Correct:

The activity and assigned scope do not match the policy, even though the sensitive information type matches.
A DLP policy fires only when its configured scope, detection type, and monitored activity or location all apply to the event. Matching the sensitive information type alone is insufficient when the event occurs on a different surface or outside the assigned scope.

Wrong:

The policy cannot detect credit card numbers when they appear in AI-generated messages.
Credit card numbers can be used as a DLP detection condition for appropriately configured AI-related activity. The issue is whether the policy is monitoring the relevant activity and scope.

The policy evaluates the message but suppresses the alert because an AI agent sent it.
A policy configured only for endpoint file-upload activity does not evaluate agent-to-human messaging as though it were an endpoint upload.

The policy requires the credit card number to appear in both an uploaded file and an agent message.
The policy does not require the same sensitive data to appear across two different activities. Each policy evaluates events that fall within its configured activity or location.