Microsoft Cloud and AI Security Engineer Associate (SC-500)

B is correct.

Explanation: An organization-level connector connects the AWS management account. Defender for Cloud uses AWS StackSets to automatically create connectors for all child accounts, including accounts added later. This gives the central security team consistent coverage without creating and maintaining individual connectors. Account-level connectors fit pilots or accounts with different security requirements, which is not the stated need here.

A is incorrect: Account-level connectors cover one account each, so you would create and maintain many connectors, and new child accounts would need new connectors.

C is incorrect: An account-level connector for the management account covers only that account. It does not create connectors for child accounts, which stay unmonitored.

D is incorrect: Manual Arc deployment addresses machine protection, not account coverage. EC2 instances can use autoprovisioning through SSM instead of manual deployment.

D is correct.

Explanation: Runtime container scanning evaluates images actually running in AKS clusters, including images pulled from public registries or private registries not connected to Defender for Cloud. The recommendation Azure running container images should have vulnerabilities resolved maps vulnerable images to the specific clusters and namespaces where they run, so you can trace findings to workloads.

A is incorrect: The registry recommendation covers images stored in ACR. It does not map findings to the clusters and namespaces running them.

B is incorrect: Attack path analysis requires Defender CSPM and shows attack sequences. It is not the recommendation that maps vulnerable images to running workloads.

C is incorrect: Cloud Security Explorer requires Defender CSPM to be enabled. Vulnerability findings still appear in Recommendations without it. Use the runtime recommendation.

C is correct.

Explanation: During prompt processing, Microsoft Security Copilot generates a process log that is visible to the user. Security analysts review this log to see which specific capability was chosen to generate the response. This visibility enables analysts to confirm whether information originated from a trusted source and verify that required safety checks were completed successfully.

A is incorrect: The orchestrator determines context and builds execution plans internally, but it does not provide analysts with a separate backend planning summary.

B is incorrect: The prompt bar serves only as the interface where users submit natural language prompts, rather than displaying capability execution or source trustworthiness.

D is incorrect: Copilot evaluates plugin data while executing its plan, but it does not supply raw pre-execution plugin telemetry for reviewing selected response capabilities.

A is correct.

Explanation: A trust policy mismatch happens when manual edits to the CloudFormation-created roles break the trust policy. Rerunning the CloudFormation template restores the original trust conditions, which lets Defender for Cloud assume the role through web identity federation again. Open Environment settings after the template runs to confirm that the Connectivity status returns to Healthy.

B is incorrect: Scan interval controls how often Defender for Cloud polls AWS APIs. It does not repair a broken IAM role trust policy.

C is incorrect: The SSM Agent is needed for Arc provisioning on EC2 instances. It does not fix the role trust policy.

D is incorrect: Adding regions extends coverage to unselected regions. It does not restore the trust conditions on the roles that were modified in IAM.

B is correct.

Explanation: The Data and AI security dashboard requires multiple plans for complete visibility. While the AI sections populate from the AI services plan, the data sections are populated by Defender CSPM with the sensitive data discovery extension enabled, Defender for Storage, and Defender for Databases. Enabling those specific workload plans populates the missing data sections.

A is incorrect: Security reader is the minimum required role to view the dashboard. Assigning the Owner role grants unnecessary management permissions without populating underlying resource telemetry.

C is incorrect: Suspicious prompt evidence provides redacted prompt and response text inside AI alert details. It does not provide discovery telemetry for databases or storage account resources.

D is incorrect: AI model security scans models registered in Azure Machine Learning registries for vulnerabilities. It is an AI component and cannot populate dashboard data resource metrics.

A is correct.

Explanation: On the Microsoft 365 connector page within the Instructions tab, the Configuration section provides checkboxes for Exchange, SharePoint, and Teams. To meet the specific organizational requirement, the analyst marks the checkboxes for Exchange and Teams while leaving SharePoint cleared, then selects Apply Changes and waits until validation completes and the button changes to Disconnect.

B is incorrect: The connector configuration requires selecting specific record types under Configuration and choosing Apply Changes, not selecting Connect from the preview pane.

C is incorrect: The Azure Policy Assignment Wizard is used specifically for the Azure Activity connector to configure subscription logs, not Microsoft 365.

D is incorrect: The Create Incidents option belongs to the Microsoft Entra ID Protection connector, which sends alert data into the SecurityAlert table.

C is correct.

Explanation: To use TLS communication between the Syslog source and the Syslog forwarder, configure the Syslog daemon on the forwarder, either rsyslog or syslog-ng, to communicate in TLS. The daemon manages inbound communication before AMA forwards collected logs. Port and network adjustments alone do not establish encryption, making daemon configuration the necessary action.

A is incorrect: Changing daemon ports and protocols aligns the forwarder with network policies, but it does not configure the daemon to communicate using TLS.

B is incorrect: Reinstalling AMA does not configure TLS between the Syslog source and forwarder, as transport encryption is handled by the Syslog daemon.

D is incorrect: Aligning the forwarder network with corporate network security policies satisfies network rules, but it does not enable TLS encryption on Syslog.

B is correct.

Explanation: In Advanced Hunting, AIAgentsInfo is the dedicated table for agent inventory and posture assessment. It logs configuration details for each discovered agent, including tools, knowledge sources, authentication settings, and administrator block status. Security analysts specifically query AIAgentsInfo to evaluate posture risks, whereas the CloudAppEvents table tracks runtime actions, tool invocations, and data access events.

A is incorrect: Inspecting what an agent did at runtime, such as data access events or tool invocations, requires querying CloudAppEvents rather than AIAgentsInfo.

C is incorrect: Correlating blocked interactions with identity and endpoint signals occurs in Defender XDR Incidents, not by constructing KQL queries in AIAgentsInfo.

D is incorrect: Checking maker protection status is done on the Agents page within Copilot Studio, not through Advanced Hunting queries in AIAgentsInfo.

D is correct.

Explanation: Alerts flow from Microsoft Defender XDR to Microsoft Sentinel through the Microsoft Defender XDR data connector. Because the AI alerts already appear in the Microsoft Defender portal, detection and delivery to Defender XDR are functioning properly. Confirming that the data connector is active is the necessary step to restore incident synchronization into Microsoft Sentinel.

A is incorrect: Suspicious prompt evidence adds redacted prompt and response snippets to alert details. It affects alert context during triage but does not control alert ingestion into Sentinel.

B is incorrect: Email notification settings determine recipient email addresses and severity filters for subscription notifications. They operate independently of incident forwarding to the Microsoft Sentinel workspace.

C is incorrect: Data security for AI interactions shares prompt metadata with Microsoft Purview for compliance tasks. It does not manage how Defender for Cloud alerts flow into Microsoft Sentinel.

C is correct.

Explanation: Agent user account creation appears as a Create user audit event, not as a service principal event. To monitor both agent identities and agent user accounts, filter for Add service principal and Create user. Then verify the object type, because the events do not identify agents by themselves. The Add service principal filter alone misses that creation event.

A is incorrect: Update service principal events show changes to agent configuration or permissions. They do not record the creation of an agent user account.

B is incorrect: App role assignment events show access granted to an agent, not the creation of an agent user account in this scenario.

D is incorrect: Delete service principal events show removed agent identities, so they cannot reveal a newly created agent user account in the audit log.