Securing AI Agents in Microsoft Foundry with Microsoft Security
Microsoft Events presents an advanced Microsoft Ignite session led by Neta Haiby, Shravana Mukherjee, and Peter Simones, focusing on securing AI agents in Microsoft Foundry using Microsoft Security tools and governance techniques.
Securing AI Agents in Microsoft Foundry with Microsoft Security
Session Overview
In this advanced breakout session from Microsoft Ignite 2025, security experts discuss ways to safeguard AI agents built in Microsoft Foundry. The session highlights essential security interventions spanning the AI agent lifecycle, with practical demos using Microsoft Security tools.
Key Topics
- AI Threat Landscape Mapping: Utilizes MITRE, OWASP, and Atlas frameworks to assess risks.
- Usage Layer Risks: Addresses insider threats and overreliance problems associated with AI agents.
- Data Labeling and Access: Shows how access to sensitive data can be controlled for different audiences including executives and entry-level employees.
- Agent Runtime Interventions: Explains six critical intervention points to improve agent security, from authentication to ongoing policy enforcement.
- Identity Management: Demonstrates the Microsoft Intra Agent ID (Public Preview), enhancing identity and access management for AI agents.
- Conditional Access Policies: Step-by-step demo of setting up Conditional Access policies via the Agent ID interface.
- Application Components and Security Points: Overview of security checkpoints at various stages of AI application development and deployment.
- Defender Integration: Demonstrates Defender AI Agents tab and recommendations for protecting AI platforms.
Demonstrations & Implementation
- Agent ID Interface: Setting up and managing agent identities and policies.
- Conditional Access Setup: Walkthrough configuring access based on organizational roles and data labels.
- Defender AI Recommendations: Insights on agent posture, recommendations, and monitoring features.
Resources
Speakers
- Neta Haiby
- Shravana Mukherjee
- Peter Simones
This session provides actionable strategies for Microsoft consultants and developers to secure AI solutions developed in Microsoft Foundry, leveraging current Microsoft Security products.