Portal Walkthrough: Control 1.24 - Defender AI Security Posture Management (AI-SPM)
Last Updated: January 2026 Portal: Microsoft Defender for Cloud (Azure Portal) Estimated Time: 2-3 hours
Prerequisites
- Azure subscription with Defender for Cloud enabled
- Security Admin or Subscription Owner role
- AI workloads deployed (Azure AI Foundry, Azure OpenAI, or Copilot Studio with Azure integration)
Step-by-Step Configuration
Step 1: Enable Defender for Cloud
- Open Azure Portal
- Navigate to Microsoft Defender for Cloud
- Select Environment settings from the left menu
- Select your subscription
- Ensure Defender CSPM plan is enabled (required for AI-SPM)
Step 2: Enable AI Security Posture Management
- In Defender for Cloud, navigate to Environment settings
- Select your subscription
- Under Defender CSPM, click Settings
- Enable AI security posture management toggle
- Click Save
Step 3: Configure AI Workload Discovery
- Navigate to Inventory in Defender for Cloud
- Filter by resource type: AI/ML services
- Verify AI resources are discovered:
- Azure OpenAI Service
- Azure AI Services (Cognitive Services)
- Azure Machine Learning workspaces
- Azure AI Foundry projects
Step 4: Review Attack Paths
- Navigate to Attack path analysis in Defender for Cloud
- Filter by AI workloads or search for "AI"
- Review attack paths targeting:
- AI model endpoints
- Data stores used by AI services
- Service principals with AI permissions
- Prioritize remediation based on risk score
Step 5: Review AI Security Recommendations
- Navigate to Recommendations in Defender for Cloud
- Filter by:
- Resource type: AI/ML services
- Category: Identity, Data, Network
- Review recommendations such as:
- Enable managed identity for AI services
- Restrict network access to AI endpoints
- Enable diagnostic logging for AI services
- Rotate API keys regularly
Step 6: Configure Multi-Cloud Connectors (Optional)
If you have AI workloads in AWS or GCP:
- Navigate to Environment settings
- Click Add environment > Amazon Web Services or Google Cloud Platform
- Follow the connector wizard
- Enable AI workload discovery for:
- AWS: Amazon Bedrock, SageMaker
- GCP: Vertex AI, AI Platform
Step 7: Set Up Alerting
- Navigate to Security alerts > Alert rules
- Create custom rules for AI-specific scenarios:
- Unusual AI API access patterns
- High-volume prompt requests
- AI model configuration changes
- Configure notification to SOC team
Configuration by Governance Level
| Setting | Baseline (Zone 1) | Recommended (Zone 2) | Regulated (Zone 3) |
|---|---|---|---|
| AI-SPM Enabled | Yes | Yes | Yes |
| Discovery Frequency | Daily | Daily | Continuous |
| Attack Path Review | Monthly | Weekly | Daily |
| Recommendation Review | Monthly | Weekly | Daily |
| Critical Remediation SLA | 30 days | 14 days | 72 hours |
| Multi-cloud Connectors | Optional | If applicable | Required if applicable |
| Sentinel Integration | Optional | Recommended | Required |
Validation
After completing these steps, verify:
- AI-SPM is enabled in Defender for Cloud
- AI workloads appear in inventory
- Attack paths targeting AI are visible
- Security recommendations are generated
- Alerting is configured for AI security events
Back to Control 1.24 | PowerShell Setup | Verification Testing | Troubleshooting