san wrote: > Hello sir, > > I need your help to avoid this type of spam. i tried SARE rules but not able > find any score. Can anyone write a small rule. It will be great help for me > and iam still learning spamassassin and not able to catch this one. > http://www.nabble.com/file/4147/States.gif > http://www.nabble.com/file/4148/s.txt s.txt > That's part of the new "image spam" wave.
There are several things that help, but none is perfect. 1) The SARE stocks ruleset. Be sure to update it frequently, or use RDJ, as this spam pattern changes a lot. Most of the image spams are stock ads, so the rules have been going here for now. 2) Many of the senders seem to be listed in the spamhaus XBL list. Make sure you're using RBLs with SA. 3) Bayes training with sa-learn --spam. Minimally effective, but at least it keeps these at BAYES_50 or so, instead of BAYES_00. 4) add one or more of the image processing plugins. ImageInfo is the lightest-weight. FuzzyOCR uses an OCR on the images and can be pretty expensive, but helps with the ones that don't have messy backgrounds. You can also take other more drastic measures like greylisting, but these have drawbacks of their own.
