RBL: SpamCannibal is a Honeypot. The only way to get listed on SpamCannibal is for a Mail Server on an IP Address to send their Honeypot unsolicited spam. The problem is when an email account gets compromised (due to an insecure, brute-forced, or stolen password) and sends out spam. It doesn't matter if it is GMail address or a Comcast.net address, everyone who uses that same Mail Server on that IP Address to send email gets flagged by the SpamCannibal RBL.
If you use SpamCannibal and are getting a large number false positives I would recommend decreasing the Weight and/or removing the "Enable for Incoming SMTP Blocking" (you would still want to "Enable for Filtering". That way your users will still be able to receive emails flagged by SpamCannibal (they won't be blocked entirely) but depending on the overall weight of the email it may be filtered according to their Spam Filtering actions.
Personally we don't use RBL: SpamCannibal on our Mail Servers. I think we were getting 0 hits from it during one of our audits (performed over the course of 14 days) and disabled it because it wasn't an effective means for us (we've disabled 29 RBLs/URIBLs in Smartermail and added 32 others to replace them over the years).