Bayes Poisoning? Spam with negative BAYES Scores

Pete pete at
Fri Dec 19 02:53:11 GMT 2003

Ken Anderson wrote:

> Just .02..
> For most of these, the bayes poisoning seems to be done in the
> text/plain portion of the email, while the html part is spammy.
> SA should be able to ignore the text/plain part completely if there is a
> spammy html version. I'm not sure what/if anything has been done along
> these lines.
> Ken A.
> Pacific.Net
> Nathan Johanson wrote:
>> Hello,
>> A recent thread made brief reference to "bayes poisoning". We're
>> currently using SpamAssassin 2.60 and will upgrade to version 2.61
>> shortly, but we have seen an increasing amount of spam slipping through
>> the filters with negative BAYES scores. Very annoying:
>> X-tcpnetworks-MailScanner: Found to be clean
>> X-tcpnetworks-MailScanner-SpamCheck: not spam, SpamAssassin
>> (score=-3.361,
>>  required 4, BAYES_00 -4.90, HTML_IMAGE_ONLY_06 1.44,
>>  HTML_MESSAGE 0.10)
>> Is anyone else seeing this sort of thing? Any recommendations for
>> combating it? I noted the links to BigEvil custom rules (posted
>> earlier), but I'm starting to wonder if Bayes usefulness is starting to
>> dwindle. Is there some way to prevent this poisoning?
>> help too.
>> Nathan
I am starting to find that as the bayes DB is getting larger that more
spam is starting to get through. I have only installed 6 weeks ago and
in the last 2 weeks i have a steady increase in spam not being trapped -
is there bayes maintenance i need to do? maybe its something completely
unrelated, but it seemed logical to me.

More information about the MailScanner mailing list