Important improvements in the false-positive rate of the outbound spam classification [Ticket #:16175857]
I reported some issues concerning Office 365 outbound spam detection to Microsoft under [Ticket #:16175857].
Outbound spam detected mails are sent through a high-risk delivery pool (HRDP) and hence are highly probable of not being accepted by business partners.
Here are our improvement suggestions:
At the moment, forwarded emails that contain links pointing to safelinks.protection.outlook.com from another Microsoft client who uses the SafeLinks feature will be classified by Microsoft's Exchange Online as outbound spam, if replied to or otherwise forwarded.
This behavior should be changed. Links that are manipulated by Microsoft's own SafeLinks feature should not be classified as outbound spam by Microsoft's Exchange Online servers at another Office 365 customer.
At the moment, emails that contain just one phrase from Microsoft's spam dictionary cause mails to be classified by Microsoft's Exchange Online as outbound spam.
One example is the following prase: replica watch
Just these two words somewhere in an unsuspicious business email suffice to classify a mail as outbound spam.
This behavior should be changed. The outbound spam detection should be more sophisticated and consider context and/or other clues.
sound good pls do it