As of 7.6.3, Anti-Spam settings > Spam Terms do not seem to be adding any additional score. I've tried scores of 5 and 100 and have tried with normal or score based dictionaries. The spam dictionaries work properly when used in compliance rules within the same policy. Anyone else experience this? Here is an example below:
I haven't seen any instances of this. If you are seeing this not working properly, I would recommend opening up a ticket with support to get this looked into and see if there is a misconfiguration or if perhaps it's a bug or something. I would recommend that enabling the XML channel logging and sending some test messages which should get tagged would probably be a good idea.