Question 410
A Graph script changes compliance policy assignments in production without recording input or output. Which finding would support closing the deployment or security issue?
Explanation
Correct Answer
This is the acceptance evidence for this specific situation: Every actual assignment update is attributable to an approved request and measured result. It would confirm that the intended intervention, 'Add dry-run mode, explicit change approval, logging and post-change verification.', addressed the problem because high-impact automation needs auditability and rollback evidence.
Incorrect Answers
This observation would validate a different case: A support team uses Security Copilot to summarize device performance but mistakes an anomaly prediction for a completed repair. It would check 'After an approved fix, endpoint metrics improve and the recorded work item shows the change.' after 'Treat AI device performance analysis as a recommendation requiring verified action and measurable outcome.'. It does not establish success for the stated scenario; the needed evidence is 'Every actual assignment update is attributable to an approved request and measured result.'.
This observation would validate a different case: A custom compliance rule treats an unavailable service as compliant because its PowerShell check returns a default true value. It would check 'A simulated missing service yields the intended noncompliant result with diagnostic context.' after 'Implement a fail-safe discovery result and explicit error handling for missing data.'. It does not establish success for the stated scenario; the needed evidence is 'Every actual assignment update is attributable to an approved request and measured result.'.
This observation would validate a different case: A PowerShell policy script updates 200 devices without checking that the device inventory changed since its last read. It would check 'The dry run rejects stale selections and records the updated approved device set.' after 'Add preflight freshness checks and compare identifiers before applying bulk changes.'. It does not establish success for the stated scenario; the needed evidence is 'Every actual assignment update is attributable to an approved request and measured result.'.