Skip to content
Submit incident
AII-1001-08391
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.Supporting sources are pending verification.

Microsoft 365 Copilot Bug Exposed Confidential Emails to AI

Microsoft confirmed a bug in its 365 Copilot allowed the AI system to read and summarize confidential emails marked with data loss prevention labels without authorization from January through early February 2026. The bug affected business and enterprise users and bypassed security controls intended to prevent sensitive workplace data from being processed by the AI system.

January 2026 Β· Microsoft Β· Privacy and surveillance
January 2026
AII-1001-08391
Quick facts
Category
Privacy and surveillance
Country
United States
Date
January 2026
Drafted by
Raidu Incident Detection

What happened

Microsoft confirmed a bug in Microsoft 365 Copilot allowed Copilot Chat to read and summarize users' confidential emails without consent starting in January 2026.

The bug affected draft and sent email messages with confidentiality labels applied, which should have been blocked by data loss prevention policies, but were incorrectly processed by Copilot Chat integrated into Outlook, Word, Excel and PowerPoint.

Microsoft began rolling out a fix in early February 2026, tracked internally as service issue CW1226324.

Reported impact

The bug mainly impacted business and enterprise users of Microsoft 365 Copilot features at work.

Microsoft did not disclose how many customers were affected, whether email content was stored beyond generating summaries, or if the issue impacted all regions or only certain enterprise environments.

Index analysis

The bug bypassed data loss prevention controls intended to prevent automated ingestion of sensitive information into large language models.

Copilot Chat's deep integration with workplace data to provide context-aware responses created conditions where small bugs could raise serious privacy concerns.

Governance controls

Microsoft said admins can monitor the issue through its service dashboard.

The incident raised concerns about how AI tools handle sensitive workplace data, with some organizations already restricting built-in AI features on official devices over data security worries.

Classification

Category
Privacy and surveillance
Industry
Digital security
Country
United States
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harmed
Business, Consumers
Failure modes
Agent misaction, Data leak, Content moderation failure
AI systems
Copilot Chat, Microsoft 365 Copilot

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident