The difficulty I forsee is that all most spam is likely to be unencrypted. Are you going to be charging the sender or the recipient for those messages?
Well, now that you posed that problem, automatically-detected spam will not be charged. Note that e. g. SpamAssasin usually makes false positives and not false negatives, so the effect will be that some legitimate plaintext messages will not be charged.
However, I would like to inform you that my SecuriNet uses completely-different technologies to eliminate spam. I haven't been able to make it, yet, due to repetitive government attacks against me. For this project, I will mostly use common technology and not my SecuriNet and related technologies.