- Microsoft marked some legitimate adobe -e emails like spam
- The company accused it for a mistake with its machine learning model
- The problem now seems to be resolved
Microsoft has corrected an annoying error in which a machine learning model, in turn, was erroneously marked with legitimate Adobe -e emails like spam.
The erroneous spam marking was relatively short-lived and spans a two-day period between April 22 at. 9:04 UTC and April 24 at 1 p.m. 11:04 UTC.
It is assumed that the company’s machine learning model incorrectly marked e emails because of their resemblance to spam -e emails, and especially affected e emails containing Adobe URLs.
Adobe -e -Mails are no longer marked as spam
“We have determined our machine learning (ML) model that protects exchange online from risky e -mail messages, identified incorrect legitimate e -mail messages such as spam because of their resemblance to E -mail messages used in spam attacks, resulting in influence,” Microsoft explained.
Wool Bleeping computer).
Coincident with Microsoft’s erroneous marking, recorded Malware Analysis Service Any.Run A sharp increase in the number of Adobe Acrobat Cloud -Links for personal documents submitted by Microsoft Defender XDR.
Any.Run shared on X: “After research we have discovered that Microsoft Defender XDR erroneously marked Acrobat[.]Adobe[.]Com/ID/URN: AAID: SC: As malicious. “
“To solve the problem, we started repeated Replay Time Travel (RTT) on the affected URLs to fully remedy the impact. The effect was specific to some users served through the affected infrastructure,” Microsoft added.
Any. Run noted the unintended consequences created by a combination of its services, Defender XDR and Microsoft’s Machine Learning error. Because legitimate URLs that linked to personal and working documents were submitted to someone.Run, the service users “upload more than a thousand Adobe files with sensitive choir prorate data from hundreds of businesses.”
“To stop leaks, we do all these analyzes private,” everyone confirmed. Run.
