Microsoft 365 Copilot Prompt Injection Threats in Enterprise
Microsoft has released a structured investigator playbook for reconstructing AI-related activity across Microsoft 365 Copilot and Azure AI services, addressing the challenge of converting raw telemetry into coherent incident timelines. The playbook targets threats already observed in enterprise deployments, including prompt injection attempts and unauthorized data access, and operationalizes a scope–context–signal methodology across Purview, Defender, and Sentinel. This guidance directly supports security teams responding to AI-specific incidents where unstructured telemetry has previously hindered attribution and impact assessment.
