Domain 4: Azure AI Security Controls

Microsoft SC-500 Security Copilot · 50 questions

• An Azure AI engineer needs to deploy Azure OpenAI Service so that it is not accessible over the public internet. All traffic must travel over the Microsoft backbone network from the application tier. Which network control achieves this?
• A company stores Azure OpenAI Service API keys in Azure Key Vault. An automated pipeline needs to retrieve the key at runtime. The pipeline runs on an Azure Container Apps environment. What is the most secure method to grant the container app access to Key Vault?
• An organization requires that all data stored in Azure AI Search be encrypted with keys managed by the organization rather than Microsoft-managed keys. Which Azure feature provides this capability?
• A security engineer needs to enforce that only identities within a specific Microsoft Entra ID tenant can access an Azure OpenAI resource. The organization wants to prevent cross-tenant token misuse. Which Azure OpenAI network and identity control should be implemented?
• A developer must grant a data science team read and write access to datasets in Azure Machine Learning but must not allow them to create or delete compute clusters. Which built-in Azure RBAC role assignment is most appropriate at the Azure Machine Learning workspace level?
• An organization wants to prevent Azure AI services (Cognitive Services) resources from being deployed outside of approved Azure regions. The governance policy should also prevent deploying AI resources without enabling private endpoints. Which Azure service enforces these requirements at scale?
• A security team needs to implement network segmentation for Azure AI services to ensure that Azure AI Search, Azure OpenAI, and Azure Machine Learning can communicate with each other privately, while remaining isolated from the internet. Which combination of controls achieves this?
• A company wants to enforce that Azure AI Content Safety content filtering is enabled and configured to block high-severity content for all Azure OpenAI deployments across the organization. Which approach enforces this at scale?
• A security engineer is hardening an Azure OpenAI resource and wants to ensure that if someone gains access to the Azure portal and attempts to delete the resource, they cannot do so without additional approval. Which Azure feature prevents accidental or unauthorized deletion?
• An Azure Machine Learning workspace must authenticate to an Azure Container Registry (ACR) to pull custom Docker images for compute environments. The security team requires that no service principal secrets are used. Which authentication method should be configured?
• An organization needs to configure Azure AI Services to use a virtual network service endpoint and restrict access to resources only within a specific subnet. After enabling the service endpoint, requests from on-premises networks connected via ExpressRoute should still be permitted. Which configuration supports this topology?
• A security team needs to audit all access to Azure Key Vault secrets used by Azure AI services. They want to generate alerts when a secret is accessed more than 100 times within one minute, which may indicate unauthorized automated secret scraping. Which Azure service combination achieves this monitoring and alerting?
• An organization is deploying an Azure OpenAI model to internal users and wants to ensure that all content filtering settings cannot be modified by the AI development team once set. Only the security team should be able to change content filter configurations. Which RBAC approach enforces this?
• A company deploys Azure AI Foundry for an internal AI application development team. The security team needs to ensure the team can create and manage AI projects within a hub but cannot modify the hub's network settings or add new connected resources. Which Azure AI Foundry role assignment is correct?
• A security engineer is configuring Azure OpenAI Service and needs to enforce that all connections use TLS 1.2 or higher. The organization also wants to disable older cipher suites. Which Azure feature controls the minimum TLS version for Azure OpenAI Service?
• An organization deploys Azure API Management (APIM) as an AI gateway in front of Azure OpenAI Service. The security team wants to enforce that no single application can consume more than 60,000 tokens per minute (TPM) across all Azure OpenAI calls routed through APIM. Which APIM policy implements this token-based rate limiting?
• A security engineer needs to implement a rotation policy for the customer-managed key (CMK) used to encrypt an Azure AI Search index. The organization requires key rotation every 90 days and automatic disabling of old key versions after 180 days. Which Azure Key Vault feature automates this lifecycle?
• An organization needs to grant a specific application identity the ability to call Azure OpenAI's chat completions API and submit batch jobs, but not the ability to view or modify resource configurations, deployments, or quotas. Which RBAC configuration achieves this least-privilege access?
• A security team wants to implement network isolation for Azure Machine Learning compute instances to prevent them from accessing the public internet while still allowing them to communicate with Azure Container Registry and Azure Blob Storage. Which Azure ML workspace configuration achieves this?
• An organization uses Azure OpenAI Service with content filtering and needs to configure the content filter to allow medical and clinical content that would normally be blocked at the default severity settings (because it contains descriptions of injuries and medication effects for a legitimate clinical decision support tool). Which approach allows this content type while maintaining safety controls?
• A company's security team wants to ensure that all Azure AI services in the organization's subscriptions have diagnostic settings configured to send logs to a central Log Analytics workspace. Which Azure tool enforces this requirement across all current and future AI service deployments?
• An organization is deploying Azure AI Foundry with a private hub that uses private endpoints. The development team reports that they can access the Azure AI Foundry portal from within the VNet but cannot access it from their home offices. The security team needs to enable remote access without exposing the hub to the public internet. Which solution achieves this?
• A security engineer wants to configure an Azure Monitor alert that fires whenever the latency of Azure OpenAI API responses exceeds 10 seconds for more than 5 consecutive minutes. This would indicate a potential denial-of-service condition or resource exhaustion. Which Azure Monitor resource provides the Azure OpenAI latency metric?
• A Defender for Cloud recommendation shows that an Azure AI Services resource has 'Managed identity not enabled,' rated as medium severity. The security team wants to remediate this across 20 Azure AI Services resources at once. Which Defender for Cloud feature enables bulk remediation?
• A security engineer is deploying an Azure OpenAI service for a healthcare application that processes PHI. The requirement mandates that the service must not use Microsoft-managed keys and all model data must be encrypted with customer-managed keys. What must be configured?
• A security architect needs to deploy an Azure AI Foundry workspace so that all training compute and inference endpoints are accessible only from within the corporate virtual network. No public endpoint exposure is acceptable. Which combination of configurations achieves this?
• A developer asks a security engineer why their Azure AI services application is getting 403 errors when using a managed identity to call Azure OpenAI. The managed identity is assigned the 'Cognitive Services OpenAI User' role on the Azure OpenAI resource. What is the most likely cause?
• A security engineer is hardening an Azure AI Search instance used in a RAG pipeline. The requirement is that data indexed by Azure AI Search must be encrypted at rest using a key that can be revoked instantly if a breach is detected. What must be configured?
• An organization needs to implement network isolation for an Azure AI Foundry hub deployed in a managed virtual network. The AI models need to access an Azure Storage account but all access must remain private. What should be configured?
• A security engineer is reviewing the access control configuration for an Azure Machine Learning workspace. Multiple data scientists need to run experiments and deploy models but should not be able to modify workspace-level settings or manage other users' permissions. Which built-in role should be assigned?
• An organization's AI application stores conversation history in Azure Cosmos DB. A security engineer must ensure the application's managed identity can only read conversation data and cannot write or delete records. Which approach achieves least-privilege access to Cosmos DB?
• A company deploys an AI chatbot on Azure that uses Retrieval-Augmented Generation. The security team wants to implement output validation to prevent the chatbot from responding with content that contradicts the organization's official policies. Which Azure AI Foundry capability enables this output guardrail?
• An organization wants to use Azure AI Language service for sentiment analysis of customer feedback. The security team requires that the service is accessed via a private endpoint and API keys are stored securely. The application runs in Azure App Service with a managed identity. What is the recommended secure configuration?
• A security engineer must configure Azure Policy to enforce that all Azure AI services (Cognitive Services, Azure OpenAI, Azure AI Search) in the organization must disable public network access. Which Azure Policy effect should be used to enforce this requirement on existing and new resources?
• A developer team needs to access an Azure OpenAI model deployment for development and testing. The security team wants to grant time-limited access that automatically expires after 8 hours without requiring manual revocation. Which approach achieves this?
• A security architect is reviewing an Azure AI Foundry deployment. The AI hub uses a user-assigned managed identity. The architect discovers the managed identity has 'Owner' rights on the resource group. What is the security risk and recommended remediation?
• An organization deploys an Azure Machine Learning workspace with a compute cluster for model training. The security team requires that all outbound traffic from training jobs must be inspected by Azure Firewall. How should this network architecture be configured?
• An organization needs to implement continuous compliance monitoring for all Azure AI resources against a custom compliance framework. The framework requires that all Azure OpenAI resources must have diagnostic logging enabled, use private endpoints, and be encrypted with customer-managed keys. Which Azure service provides continuous compliance monitoring and remediation for these requirements?
• An organization is deploying a large language model inference endpoint using Azure Machine Learning managed online endpoints. The security team requires that inference requests containing PII must be detected and logged separately for compliance purposes. What is the recommended approach to implement this within the Azure ML inference pipeline?
• A security engineer is deploying Azure AI Speech service for a customer call center transcription application. The transcribed audio contains sensitive customer PII. The security team requires that transcriptions are stored in a customer-owned storage account rather than Microsoft's storage. Which Azure AI Speech configuration achieves this?
• A security engineer needs to configure an Azure AI Language resource used for document analysis to accept requests only from a specific Azure Virtual Network subnet. Which two resources must be configured to achieve this network restriction?
• A company builds a chatbot using Azure OpenAI with a system prompt containing proprietary business logic. The security team is concerned that users could extract this system prompt through carefully crafted questions. Beyond Prompt Shield, which Azure OpenAI feature can be configured to add a layer of protection against system prompt extraction?
• An organization is implementing Azure Machine Learning with sensitive training data in Azure Data Lake Storage Gen2. The security team requires that the AML workspace's managed identity can only access specific containers in the Data Lake, not the entire storage account. How should this access be configured?
• A security engineer is configuring Azure Content Safety for a customer-facing AI application. The application serves users in multiple regions and must comply with regional content regulations. The engineer needs to configure content filtering thresholds that are stricter for the EU region compared to the US region. How should this be implemented?
• An organization wants to implement a proactive security control for their Azure OpenAI deployment that automatically blocks known malicious IP addresses identified by Microsoft threat intelligence. Which Azure native service enforces this at the network level?
• A security engineer is reviewing the access control model for a multi-model Azure AI Foundry hub that manages separate AI projects for the HR team and the Finance team. The engineer needs to ensure that HR project data scientists cannot access Finance project models and vice versa, while allowing the platform admin to manage all projects. Which Azure AI Foundry access control configuration achieves this?
• A security engineer is integrating Microsoft Defender for Cloud recommendations for Azure AI services into the organization's vulnerability management program. Defender for Cloud shows a high-severity recommendation: 'Azure AI services resources should have key access disabled (disable local authentication).' What is the security benefit of implementing this recommendation?
• A security engineer at a company using Azure OpenAI for document analysis needs to ensure that the application properly handles authentication token expiration without service disruption. The application uses a managed identity. Which Azure SDK pattern implements proper token lifecycle management?
• An organization deploys Azure AI Content Safety to moderate user inputs to a consumer-facing AI application. The content safety API returns categories including Hate, Violence, Sexual, and SelfHarm with severity levels 0-6. The security team must configure the application to block any content with Violence severity 4 or higher while allowing Violence severity 0-2 for general users, but allow all violence categories up to severity 6 for verified security researchers. How should this be implemented?
• A security engineer is implementing a Bring Your Own Key (BYOK) strategy for an Azure AI Foundry deployment. The customer-managed key is stored in Azure Key Vault. The security team requires that the key must be rotated every 90 days. Which configuration ensures automatic key rotation without service disruption to the AI Foundry workspace?