Microsoft Unleashes New Copilot Privacy Control

Microsoft Purview, a robust data governance, security, and management platform, empowers enterprises to gain deeper insight and tighter control over their vast data repositories. As artificial intelligence permeates every facet of business operations, many organizations grow increasingly wary of AI systems accessing sensitive information. To assuage these concerns, Microsoft is rolling out a pivotal configuration change that enhances data protection for its commercial clientele.

Presently, users can apply Purview sensitivity labels—such as “Confidential”—to files, which triggers encryption and a visible watermark. This simple yet effective measure signals to recipients that the document is restricted and must be handled with care.

Until now, Microsoft permitted certain connected experiences—including its AI services—to analyze files regardless of their sensitivity labels. This loophole raised alarms, especially after incidents where confidential emails protected by data‑loss‑prevention policies were inadvertently uploaded to Copilot for analysis.

In response, Microsoft is tightening its Purview sensitivity settings. The updated configuration will now enforce a blanket block on all connected services—including Copilot, Word, Excel, and PowerPoint—from analyzing files marked with sensitivity labels. The labels themselves remain unchanged, but the new restriction applies universally across all AI‑driven experiences, automatically extending to any already‑labelled documents once the update propagates.

Microsoft has urged IT administrators to notify their helpdesk and compliance teams, update internal documentation, and reassess sensitivity labels to ensure alignment with their compliance requirements.

This update, identified as MC1297982 in the Message Center, will roll out in phases, becoming generally available by the end of next month. It is restricted to commercial customers holding a license that permits use of Purview.