Microsoft Security released updated research on failure modes in agentic AI systems based on twelve months of red-team testing, identifying seven new attack vectors including supply chain compromise and goal hijacking, along with mitigation recommendations.

Learn how to investigate AI activity in Microsoft 365 Copilot and Azure AI services using a structured, telemetry-driven approach. This playbook helps security teams reconstruct events, assess data exposure, and detect potential threats faster. The post Reconstructing AI activity in investigations appeared first on Microsoft Security Blog .