Question 190
A Microsoft 365 administrator implemented the following change: Require human review and source validation for consequential AI-generated content. The business requirement was: A marketing team wants AI-generated client advice to be reviewed for errors and unfair assumptions before sending it externally. Which observed test result would best confirm that the change worked?
Explanation
Correct Answer
Responsible AI includes accountability, fairness, transparency and oversight beyond mere model output generation. A saved configuration without effective access or behavior evidence would be insufficient.
Incorrect Answers
The removed user cannot open the protected content and eventual search results reflect the new access boundary. However, this does not establish the required outcome: the proposed advice is checked against verified evidence and revised before release.
The user can view content only when their label-derived rights allow it. However, this does not establish the required outcome: the proposed advice is checked against verified evidence and revised before release.
The coworker sees the shared response but cannot open the protected document unless independent authorized SharePoint access is granted. However, this does not establish the required outcome: the proposed advice is checked against verified evidence and revised before release.