BRK348 Breakout Wed 18 Nov 16:15 PST
Microsoft Purview: From understanding sensitive data to protecting it
Aashish Ramdas, Chui Yeung
- Microsoft session page (opens in a new tab)
- Watch on-demand (opens in a new tab)
- Slides: rights-protected, available via the Microsoft session page (opens in a new tab)
Summary
From the recording / Microsoft AI summary
Microsoft figures: its 500 largest customers hold 260 billion documents, of which only 65 billion were active in 30 days. A baseline scan plus rescans up to twice a year are said to be included in E5. About 100 classifiers are ready, rolling out next year at no extra cost, and a unified classifier page is in public preview. A demo showed DLP blocking Copilot from answering using a document containing a sensitive identifier.
Session description
Microsoft's catalogue description
Secure AI starts with understanding your data. Discover how Microsoft Purview uses AI-powered classification and sensitivity labels to identify sensitive content, automate protection, and help organizations confidently scale Copilot adoption.
Demo timestamps
Microsoft's player does not support links to a specific time, so open the recording and skip to the time shown.
- Skip to 10:49
DLP policy keeping a sensitive contract out of Copilot
Starting from Copilot still having access to content, a DLP policy is added to stop Copilot sharing contracts in Copilot interactions, so the contract stays out of its responses.
- Skip to 26:42
Sensitive information type-based DLP for Copilot
Walkthrough of DLP for Copilot based directly on sensitive information types, compared with label-based grounding, and how a policy controls what data goes into Copilot (main walkthrough 33:22–34:38).