AI agents are powerful, but with great power comes new security challenges. In this episode of Forward Deployed, we dive into prompt injection attacks and the emerging threat landscape for agentic systems. Our guests, Gaya Salomon and Asaf Tzuk, explain what XPIA (Cross-Prompt Injection Attacks) means, why it matters, and how Microsoft is building layered defenses to keep agents secure. We cover real-world scenarios, built-in protections, and how admins and makers can work together to safeguard their environments.
To learn more and explore related resources:
https://aka.ms/FD-StrengthenAgentsBlog
https://aka.ms/FD-DefenderAgentsBlog
Forward Deployed