Securing Microsoft Foundry with AI Gateway, Guardrails & Defender for Cloud

As organizations increasingly deploy generative AI applications, copilots, and intelligent agents, traditional security controls are no longer enough. Security teams must understand how to govern AI traffic, prevent unsafe model interactions, continuously assess AI security posture, and detect threats targeting AI workloads.

In this session, we move deeper into the security architecture required to protect enterprise AI solutions built with Microsoft technologies.

This session will focus on three critical security capabilities:

• AI Gateway security in Microsoft Foundry
• AI guardrails and content safety controls
• Microsoft Defender for Cloud protection for AI workloads

About the Session

Enterprise AI workloads introduce new attack surfaces involving prompts, model endpoints, APIs, sensitive data, model responses, identities, and connected services.

Securing these environments requires multiple layers of protection.

During this session, participants will explore how Microsoft security technologies can help organizations:

• Govern and secure access to AI models
• Control and monitor AI traffic
• Prevent unsafe or policy-violating prompts and responses
• Protect AI applications against prompt-based attacks
• Identify configuration weaknesses affecting AI resources
• Detect suspicious runtime activity
• Investigate AI-related security incidents
• Improve the overall security posture of AI workloads

The session will demonstrate how Microsoft Foundry, Azure API Management, Microsoft Defender for Cloud, Microsoft Defender XDR, and Microsoft Entra ID can work together as part of an enterprise AI security architecture.

What You Will Learn

By the end of this session, participants should understand how to:

• Explain the role of AI Gateway in securing AI model traffic
• Configure an AI Gateway in Microsoft Foundry
• Apply access controls to AI model endpoints
• Monitor and audit AI API consumption
• Understand how guardrails protect AI model interactions
• Configure content filters and blocklists
• Use Prompt Shields to strengthen protection against malicious prompts
• Test and validate AI safety controls
• Select appropriate guardrails based on workload requirements
• Enable Microsoft Defender for Cloud protection for AI workloads
• Review AI resources through the Data & AI security dashboard
• Assess AI security posture using CSPM
• Identify security gaps affecting AI workloads
• Detect runtime attacks against AI applications
• Investigate AI security alerts using Microsoft Defender XDR
• Apply defense-in-depth principles to enterprise AI environments

AI Security Architecture We Will Explore

During the session, we will connect the different technologies into a practical defense-in-depth model:

Identity Security: Microsoft Entra ID controls who and what can access AI resources.

Gateway Security: AI Gateway controls, governs, monitors, and protects traffic to AI models.

Model Interaction Security: Microsoft Foundry guardrails evaluate prompts and responses for unsafe or prohibited interactions.

Security Posture Management: Microsoft Defender for Cloud identifies configuration weaknesses and provides security recommendations.

Runtime Protection: Cloud Workload Protection detects suspicious behavior against deployed AI workloads.

Detection and Investigation: Microsoft Defender XDR brings AI-related alerts and incidents into the broader security operations workflow.

Together, these capabilities provide multiple layers of protection across the AI workload lifecycle.

Who Should Attend?

This session is ideal for:

• Participants in the Cloud and AI Security Bootcamp
• Cloud and AI Security Engineers
• Security Architects
• Microsoft Foundry administrators
• AI Engineers and Developers
• SOC Analysts
• Microsoft Defender administrators
• DevSecOps Engineers
• Cloud Security Engineers
• Microsoft Entra administrators
• Governance, Risk, and Compliance professionals
• Professionals responsible for Responsible AI
• Professionals preparing for the SC-500 certification
• Anyone interested in securing generative AI and agentic AI environments

Why You Should Attend

AI security goes far beyond protecting a model endpoint.

Organizations need visibility and control across the complete AI ecosystem—from identities and APIs to prompts, responses, cloud configurations, runtime activity, and incident investigation.

This session will help you understand how Microsoft security technologies can be combined to establish a practical defense-in-depth architecture for enterprise AI workloads.

Whether you are building AI applications, securing Microsoft Foundry environments, working in a SOC, designing cloud security architectures, or preparing for the SC-500 Cloud and AI Security Engineer certification, this session will provide practical knowledge that you can apply to real-world environments.

Come prepared to learn, ask questions, and deepen your understanding of securing enterprise AI workloads.

Learn. Connect. Secure the Future of AI.