Praetor's antispam technology does not need constant updating because it uses the advanced technique of Bayesian statistical filtering to rate the "spamicity" of received messages.  Left at "out-of-the-box" default settings Praetor will instantly stop over 90% of all spam.

Periodic use of Praetor's "on-the-fly" training, with actual samples of non-spam messages you received that were improperly classified as spam, will improve the ability to identify non-spam email.  Praetor can stop over 95% of all spam with very few mistakes. 

In general, the idea of letting outside parties determine what is legitimate email or spam is not prudent.  This determination should be based on actual message traffic.  Even when CMS provides updates to Praetor's Bayesian token database, most of the data is derived from training on spam samples.  Only your site can train Praetor on samples of legitimate messages.