Fighting spam, phishing, and online scams at the network level

Spam continues to grow at an unbridled pace, and the behavior of spammers is becoming increasingly dynamic; to stem this tide, users and operators need new proactive techniques that distinguish spammers from legitimate senders based on robust, network-level features, that are more difficult to change than the message itself. This talk described a new technique called behavioral blacklisting that allows users and operators to track the reputation of spammers based on their dynamic behavior, which tends to be more invariant than the contents of the message or the IP address from which they are sending. Although much work remains (e.g., coping with scale, evasion attempts from senders), our studies thus far indicate that behavioral blacklisting is a promising, complementary approach to existing techniques for filtering spam and fighting online scams.