Protect data and accelerate AI adoption with Microsoft Purview | BRK346
Nick Granack, Herain Oberoi, and Kim Kunes present Microsoft Purview’s approach to data security and compliance as organizations expand their use of AI. The session connects sensitive-data discovery and protection with governance, auditing, and investigations.
AI changes the data security problem
The session opens with four AI-driven cybersecurity shifts: threats, vulnerabilities, agents, and data risks. It examines why traditional cybersecurity stacks do not scale well for AI applications and agents, and highlights three data security challenges:
- The scale of data that organizations need to understand and protect.
- Additional exposure surfaces introduced by AI applications and agents.
- Compliance risks associated with AI adoption.
Microsoft Purview’s strategy is organized around understanding sensitive data, protecting it, and supporting compliance.
Classification and labeling at AI scale
Kim Kunes discusses the challenge of classifying and protecting billions of documents. The session covers enhancements to Purview classification, auto-labeling, and sensitive information type accuracy.
These topics address the need to identify sensitive information across large data estates so that organizations can apply appropriate protections.
Data loss prevention for AI workloads
The presentation covers new Purview data loss prevention (DLP) protections for:
- Microsoft Copilot.
- AI prompts.
- Grounding data.
- Local agents.
The discussion extends data protection beyond documents to the information used in AI interactions, with sensitive-data protection across Microsoft 365 and Copilot as a central focus.
Compliance, auditing, and AI governance
The session examines Compliance Manager, auditing, and readiness for AI regulations. An EY discussion closes the presentation with practical governance topics: simplifying labeling strategies, improving visibility, and establishing audit best practices.
Session chapters
- 00:00 — Opening remarks and Microsoft Purview for AI data protection.
- 02:29 — Four AI-driven cybersecurity shifts: threats, vulnerabilities, agents, and data risks.
- 04:35 — Why traditional cybersecurity stacks do not scale for AI applications and agents.
- 05:41 — AI-driven data security challenges: scale, exposure surfaces, and compliance risks.
- 07:32 — Purview strategy: sensitive-data understanding, protection, and compliance.
- 08:24 — Classification at AI scale and protecting billions of documents.
- 11:28 — Classification, auto-labeling, and sensitive information type accuracy enhancements.
- 16:24 — DLP protections for Copilot, AI prompts, grounding data, and local agents.
- 25:06 — Compliance Manager, auditing, and AI regulatory readiness.
- 29:04 — EY discussion on AI governance, labeling strategies, visibility, and audit best practices.
This Microsoft Ignite 2026 breakout session, BRK346, is an intermediate-level security session presented in English (US).