How a Microsoft Defender XDR Error Triggered a Massive Data Leak
Security researchers at ANY.RUN recently uncovered a critical flaw in Microsoft Defender XDR, a leading threat detection platform. The incident exposed over 1,700 confidential business documents from hundreds of companies, raising urgent questions about the reliability of automated security systems.
On analyzing a sudden spike in uploads to their public sandbox, ANY.RUN traced the issue back to a misclassification error in Microsoft Defender XDR. The platform incorrectly flagged legitimate Adobe Acrobat Cloud URLs—specifically links starting with acrobat[.]adobe[.]com/id/urn:aaid:sc:—as malicious.
This false positive triggered Defender XDR’s automated response: quarantining files and redirecting them to ANY.RUN’s sandbox for analysis. However, users on the platform’s free tier, which defaults to public sharing, unknowingly exposed sensitive corporate data.
The leaked files contained:
ANY.RUN acted swiftly to mitigate damage by converting affected analyses to private mode. Despite this, some users continued uploading sensitive files publicly, compounding the breach.
Final Takeaway:
Automated tools like Microsoft Defender XDR are invaluable, but human vigilance is irreplaceable. Always pair technology with robust protocols to avoid becoming the next headline.