A Zero Trust approach to securing AI runtime with Microsoft Defender | BRK340
Krishna Sunkavalli and James Brotsos present this intermediate-level Microsoft Ignite 2026 session for teams securing AI agents at runtime.
Securing AI agents at runtime
The session treats AI agents as an attack surface and explores how Microsoft Defender helps apply a Zero Trust approach to reduce agent risk. It focuses on defending against three categories of AI attacks:
- Prompt injection
- Tool misuse
- Identity abuse
The security guidance is mapped to OWASP and MITRE, connecting AI runtime protection to these security frameworks.