SA Bugzilla – Bug 4112
rule to detect misleading hyperlinks
Last modified: 2005-02-17 15:07:55 UTC
(Sigh) I'm so sick of getting phishing emails. A common characteristic seems to be the use of misleading hyperlinks, as in this example: <a href="http://email.apollo-cn.idv.tw/webmail/database/.wamu/" >https://login.personal.wamu.com/registration/CreateLogonEntry.asp</a> A rule could be <\s*a\s+href\s*=\s*['"](.*?)['"]\s*>(https?:.*?)</\s*a\s*> where $1 != $2. But I don't know how to express that kind of condition as a simple regular expression. Please forgive me if this RFE is off-base... I'm not a spamassassin expert, just a satisfied user.
Subject: Re: New: rule to detect misleading hyperlinks Good idea, and expressiable in an regex. But it doesn't work well. There are too many legit sites that do things where there is a chanracter or two difference between the uris, or they are even completely different. This is especially bad in newsletters. This is probably a case where more specifically targeted rules will have a better chance of working. SARE has a number of anti-phishing rules that work fairly well, although they could be improved. WAMU in particular is a fairly new phishing target.
Unfortunately, this is really common in legitimate mail. Don't ask me why, but it is... I've tested this idea before quite extensively.