Secure AI solutions in the cloud Training (SC-5009)

Generative AI deployments in the cloud introduce new attack surfaces and compliance challenges. The SC-5009 Applied Skill prepares you to secure these solutions end-to-end by leveraging Microsoft Defender for Cloud and Microsoft Purview, the two pillars of Microsoft’s cloud AI security offering.

You learn to detect AI workload threats, classify and protect sensitive data consumed by AI models, and implement governance controls aligned with the Microsoft Responsible AI Standard. The training is delivered in Geneva and Lausanne by a Microsoft Certified Trainer.

Participant Profiles

Configure Microsoft Defender for Cloud to protect AI workloads and detect threats
Classify and protect sensitive data with Microsoft Purview Information Protection
Apply data lifecycle and retention policies adapted to AI use cases
Detect and respond to prompt injection, data leakage and model abuse
Align cloud AI deployment with the Microsoft Responsible AI Standard

Module 1 : Understand how Microsoft Defender for Cloud supports AI security and governance in Azure

Understand AI services in Azure
Understand AI security risks in Azure
AI guardrails and protections in Azure
How Azure security and governance tools support AI workloads
Module assessment

Module 2 : Protect AI workloads with Microsoft Defender for Cloud

Enable the AI workloads plan
Review insights in the Data & AI security dashboard
Assess and improve AI security posture with Cloud Security Posture Management (CSPM)
Detect AI threats at runtime with Cloud Workload Protection (CWP)
Investigate AI security alerts with prompt evidence in Microsoft Defender XDR
Module assessment

Module 3 : Configure and manage guardrails in Microsoft Foundry

Understand guardrails and Microsoft Content Safety
Understand safety controls in Microsoft Foundry
Try out built-in guardrails
Create and manage blocklists in Microsoft Foundry
Configure and apply guardrails in Microsoft Foundry
Choose and refine the right guardrails for your AI workloads
Module assessment

Module 4 : Secure Microsoft Foundry environments

Control access to Microsoft Foundry with Microsoft Entra ID
Manage access within Microsoft Foundry projects
Secure Microsoft Foundry secrets with Azure Key Vault (preview)
Isolate networks with managed virtual network and Private Link
Enable diagnostic logging in Microsoft Foundry
Module assessment

Module 5 : Understand identity architecture for AI workloads

Identity as the control layer for AI solutions
Management plane and data plane access in AI workloads
Authentication flows for AI endpoints in Microsoft Foundry
Human and workload identities in AI workloads
Role assignments and scope in AI environments
Common identity misconfigurations in AI deployments
Module assessment

Module 6 : Implement access management for Azure resources

Assign Azure roles
Configure custom Azure roles
Create and configure managed identities
Access Azure resources with managed identities
Analyze Azure role permissions
Configure Azure Key Vault RBAC policies
Retrieve objects from Azure Key Vault

Module 7 : Plan, implement, and administer Conditional Access

Plan security defaults
Plan Conditional Access policies
Implement Conditional Access policy controls and assignments
Test and troubleshoot Conditional Access policies
Implement application controls
Implement session management and continuous access evaluation
Microsoft Entra Conditional Access Optimization agent
Module assessment

Module 8 : Manage Microsoft Entra Identity Protection

Review identity protection basics
Implement and manage user risk policy
Monitor, investigate, and remediate elevated risky users
Implement security for workload identities
Explore Microsoft Defender for Identity
Explore the Identity Risk Management Agent
Module assessment

Course material included.

Complementary Courses

Additional Information

Why secure cloud AI solutions with Microsoft Defender for Cloud and Purview

The deployment of cloud AI solutions multiplies the security risks: exposure of sensitive data in prompts, prompt injection, jailbreak, data exfiltration via model responses. Microsoft offers a coherent answer with Defender for Cloud for runtime protection and Purview for data governance. The SC-5009 Applied Skill trains you to combine these two layers for end-to-end cloud AI security.

Microsoft Defender for Cloud for AI workloads

The course covers Defender for Cloud features dedicated to AI workloads: detection of suspicious calls to Azure OpenAI Service, monitoring of prompts and responses, alerts on unusual generation patterns, integration with Microsoft Sentinel for centralized investigation. You configure security baselines and learn to interpret AI-related security recommendations.

Microsoft Purview for data classification

The training details how to use Microsoft Purview to classify the data consumed by AI agents and assistants: sensitivity labels, retention policies, data lifecycle management. You also configure content discovery to identify sensitive content in the data sources accessible to Copilot and your custom AI agents.

Microsoft Responsible AI Standard

The Microsoft Responsible AI Standard is the framework that guides the responsible deployment of AI. The training presents the six principles (fairness, reliability and safety, privacy and security, inclusiveness, transparency, accountability) and how to operationalize them with the available Azure AI Content Safety tools.

Audience and prerequisites

The Secure AI solutions in the cloud (SC-5009) Applied Skill is for security architects, security and compliance engineers, AI engineers and DPOs who deploy AI solutions in Azure. Basic knowledge of Microsoft Defender for Cloud and Microsoft Purview is recommended.

FAQ Secure AI solutions in the cloud (SC-5009)

What’s the difference between SC-5009 and SC-200?

SC-200 is the certification for Security Operations Analyst, covering the entire SOC (Sentinel, Defender XDR, etc.). SC-5009 is a focused Applied Skill on the specific topic of cloud AI security, complementary to SC-200.

Does the course cover prompt injection and jailbreak techniques?

Yes, with the countermeasures available in Azure AI Content Safety and Defender for Cloud monitoring patterns to detect them.

Is the SC-5009 Applied Skill equivalent to a Microsoft certification?

No. An Applied Skill is a practical skills validation format, without a formal exam. For a complete Microsoft security certification, see SC-100 (Cybersecurity Architect) or SC-200.

Training: Secure AI solutions in the cloud using Microsoft Defender for Cloud and Microsoft Purview (SC-5009)