Modern Copilot solutions introduce new risks: AI agents can inadvertently access, amplify, or leak sensitive information hidden across SharePoint, Dataverse, Teams, and unstructured content. Traditional governance is no longer enough. This session provides a practical, hands-on walkthrough of applying Data Security Posture Management (DSPM) principles to the Power Platform and Copilot Studio.
We’ll explore how AI interacts with enterprise data, how to classify and label sensitive information automatically, and how to govern what copilots are allowed to see or generate. You’ll see real demonstrations using Microsoft Purview’s sensitivity labeling, AI hub, data classification, audit trails, and Copilot Studio’s runtime logs to detect and prevent risky behavior.
We will simulate a real incident where a Copilot attempts to access a hidden “Contracts” library containing financial and PII data. Step by step, we'll discover the exposure, analyze the AI activity, classify the data, apply access controls, and validate the fix using content-aware policies.
If you want to govern Copilots responsibly in an AI-first world, this session gives you the exact process and tools to do it.