Microsoft has disclosed a bug in its Microsoft 365 Copilot Chat that allowed the AI tool to read and summarize confidential emails for weeks, bypassing data loss prevention (DLP) policies designed to protect sensitive information. The flaw, tracked as CW1226324, affected the Copilot "work tab" feature, which incorrectly processed emails in users' Sent Items and Drafts folders — even when those messages carried confidentiality labels explicitly restricting access by automated tools.

Microsoft detected the issue on January 21 and acknowledged it in early February, but the vulnerability window lasted weeks before a worldwide configuration update was deployed to enterprise customers. The company stated the bug "did not provide anyone access to information they weren't already authorized to see," though security experts note that AI summarization of confidential content still represents a significant data handling violation.

Microsoft has not disclosed how many customers were affected. The incident comes as 72% of S&P 500 companies now cite AI as a material risk in regulatory filings, highlighting growing concerns about integrating generative AI into enterprise workflows without adequate safeguards.