Hello,
I believe that this behavior is a remnant of the initial configuration and has not been changed until now also by a misunderstanding of the role of antispam. 
We use "Spam Firewall" mode, receiving messages from another server, which is why they are being filtered. In fact, this has already been useful in account 
compromise situations, where spam messages were successfully triggered, but I believe that the inconvenience of having legitimate quarantined messages does not 
compensate for maintaining this configuration.
I greatly appreciate your attention and wish you success.