Black Friday sale: an extra 40% off every course until Monday, November 30.New Year sale: an extra 30% off every course until Friday, January 1.Memorial Day sale: an extra 25% off every course until Monday, May 31.SysAdmin Day sale: an extra 25% off every course until Sunday, August 1.Back to School sale: an extra 25% off every course until Sunday, August 22.Labor Day sale: an extra 25% off every course until Monday, September 6.Black Friday sale: an extra 40% off every course until Monday, November 29.New Year sale: an extra 30% off every course until Saturday, January 1.Choose your certification →

Copyright (c) 2026 MindMesh Academy. All rights reserved. This content is proprietary and may not be reproduced or distributed without permission.

3.3.2. Protecting Sensitive Data and Evaluating Compliance Gaps

💡 First Principle: For data protection, Purview treats an agent instance much like a user: include it in policies the way you'd include a person, grant it rights to labeled content explicitly, and audit what it does. The difference is that an agent never notices a block — so someone must watch the alerts on its behalf.

Protect sensitive data. The agent details Security tab > Protect sensitive data opens Purview DSPM > AI observability (the current DSPM — the classic experience doesn't support Agent 365). Purview capabilities that support Agent 365 agents include DSPM, auditing, data classification, sensitivity labels, DLP, Insider Risk Management, Communication Compliance, eDiscovery, data lifecycle management and Compliance Manager; encryption without sensitivity labels isn't supported. Details that matter:

  • When an agent instance is created, it's automatically enabled for auditing, sensitive-data detection and Compliance Manager AI assessments. For everything else, include the agent instance in policies as you would a user.
  • DLP: scope a policy to agent instances directly or to a security group that contains them. It can block or audit agent-to-human and human-to-agent interactions in Teams, OneDrive, SharePoint and email. The agent isn't aware it was blocked, so its owners must monitor.
  • Sensitivity labels: for an agent instance to read encrypted files, the files must be shared with it explicitly and the label's encryption must grant the agent VIEW and EXTRACT — "all users and groups in your organization" isn't enough. Content an Agent 365 agent creates doesn't inherit its sources' labels.
  • Insider Risk Management uses the Risky AI usage template, with agent instances included explicitly.

Evaluate compliance gaps. The Security tab > Evaluate compliance gaps opens Purview Compliance Manager when an AI baseline assessment exists for the agent. Compliance Manager's assessments for AI regulations provide regulatory templates, and agent instances are included automatically. The default Agent 365 template's Purview AI compliance assessment policy continuously evaluates agents for compliance gaps.

All of these Purview agent features require Microsoft 365 E7 or Agent 365 and aren't available in GCC.

⚠️ Exam Trap: "Grant the agent access to the Confidential label by adding all users in your organization to the label's permissions" doesn't work — the agent instance must be granted VIEW and EXTRACT explicitly. And don't assume agent output carries labels the way Copilot output does: content created by Agent 365 agents doesn't inherit source labels.

Reflection Question: An HR agent should summarize labeled policy documents for employees but must never post salary data into Teams chats. Which two Purview configurations, aimed at the agent instance, achieve that?

See how it connects
Alvin Varughese
Written byAlvin Varughese
Founder•26 professional certifications earned