Cyber Security News

Microsoft 365 Copilot Vulnerability Exposes Sensitive Emails Through AI Summaries

A security flaw in Microsoft 365 Copilot is currently causing the AI assistant to incorrectly summarize email messages protected by confidentiality sensitivity labels, essentially bypassing configured Data Loss Prevention (DLP) policies.

This vulnerability exposes potentially sensitive organizational data to unauthorized AI processing.

The issue, tracked under Microsoft reference CW1226324, was first flagged on February 4, 2026, and involves the Copilot “Work Tab” Chat feature actively summarizing emails that carry a confidential sensitivity label, even when restrictive policies are in place.

Vulnerability DetailInformation
Tracking IDCW1226324
ComponentCopilot “Work Tab” Chat
Issue TypeDLP Policy Bypass
Root CauseCode-level defect in folder processing
Affected FoldersSent Items, Drafts

Root Cause and Technical Impact

Microsoft’s investigation identified a code-level defect as the primary root cause of this exposure.

According to CSN, the flaw allows Copilot to inadvertently pick up items stored in users’ Sent Items and Draft folders, ignoring the confidentiality labels applied to those messages.

Under normal operation, sensitivity labels paired with DLP policies should prevent Copilot from accessing or processing any email flagged as confidential.

However, the bug effectively renders these controls non-functional for the affected email folders, allowing the AI to surface restricted content in chat summaries.

This failure is particularly concerning for organizations in highly regulated industries such as healthcare, finance, and government, where email confidentiality controls are compliance requirements rather than just best practices.

The NHS has flagged the incident internally as INC46740412, indicating the issue has a real-world impact on public sector users relying on Microsoft 365.

The bypass of DLP policies by an AI assistant represents a significant security gap, as it undermines the integrity of an organization’s information protection posture.

As of February 11, 2026, Microsoft began deploying a fix across affected environments and is actively reaching out to a subset of impacted users to validate the remediation.

However, the rollout has not yet reached full saturation, and the issue remains unresolved for some organizations.

Microsoft aims to provide a remediation timeline as the fix progresses. The scope of impact is broad, potentially affecting any organization with Microsoft 365 Copilot enabled and confidentiality labels configured on email.

Administrators are advised to monitor the Microsoft 365 admin center for updates under reference CW1226324 and review Copilot activity logs for anomalous access to labeled content.

Until the fix is fully deployed, security teams should consider whether to temporarily restrict Copilot access in environments handling highly sensitive email communications to prevent accidental data exposure.

Follow us on Google News, LinkedIn, and X to Get Instant Updates and Set GBH as a Preferred Source in Google

Divya

Divya is a Senior Journalist at GBhackers covering Cyber Attacks, Threats, Breaches, Vulnerabilities and other happenings in the cyber world.

Recent Posts

Microsoft Paint Vulnerability Allows Hackers to Execute Malicious Code Remotely Without User Interaction

Microsoft has announced a vulnerability in Windows Paint that could let unauthenticated attackers execute malicious…

19 minutes ago

ARTEX AI Framework Linked to Bank Breaches and Cloud Attacks Found Exposed Across Hundreds of Servers

An autonomous penetration-testing framework linked to South Korean banking breaches and cloud supply-chain attacks is…

50 minutes ago

Shopify Makes IP Allowlisting Mandatory for All Public Apps to Prevent Stolen API Token Abuse

Shopify is making IP allowlisting mandatory for all public apps to limit abuse of stolen…

59 minutes ago

OpenAI’s New GPT-6 Cyber Finds 81 High-Severity Firefox Vulnerabilities and 948 Linux Kernel Exploits

OpenAI has introduced GPT-6 Cyber, a security-focused model built to find and validate exploitable vulnerabilities…

1 hour ago

OpenMatter Network Positions Decentralized Computing as an Alternative to Data Centers for AI

Melbourne, Florida, November 5th, 2026, CyberNewswire OpenMatter Network, a pioneer in decentralized infrastructure for agentic…

13 hours ago

China-Aligned Hackers Hijack Popular Sogou Keyboard Software to Secretly Install DarkVigil Malware

A China-aligned espionage group abused the update traffic of Sogou Input Method, one of China's…

18 hours ago