
Episode 1.10 - Threat Protection for AI: Detecting and Responding to Attacks on Copilot and Agents
AI security does not stop at prevention. In this episode of TechExplained Microsoft 365 Copilot Architecture, Laura and Mark explore how organizations can detect and respond when attackers target Microsoft 365 Copilot and AI agents. They focus on three major threats: Cross-prompt injection, where hidden instructions inside documents, emails or other content can influence Copilot or an agent.
6 September 2026 22 min
