At the same time, spammers rapidly multiplied, became far more sophisticated, and began to operate through quasi-legitimate ISPs and to move operations offshore (substantially as a result of legal risks made apparent through high profile law suits such as CompuServe Incorporated vs. Cyber Promotions Inc. and Sanford Wallace in 1997).

In the mid 1990s, the methods used to slow the flow of spam were based upon crude technological capabilities, such as restricting the "mail relay" feature of Internet email servers, technology fixes that were not designed specifically to address the spam problem.

In the late 1990s, the first true anti-spam technologies emerged, such as the unsolicited bulk email filter built into the Netscape Messaging Server product. However, most messaging infrastructure products did not have such capabilities and, by 2000, it was clear that businesses would have to play a role controlling spam in order to protect their own server networks.

It was also clear at that time that the existing capabilities of messaging infrastructure software and associated products, such as email anti-virus gateways, were inadequate and that that activities of industry standards bodies were not keeping up with the problem. This situation, combined with the fact that U.S. court cases related to spam, such as the 1997 court case mentioned above, had no apparent deterrent effect on spammers, presented a potential business opportunity for anti-spam technologies.

Precursors of anti-spam technology
Virtually every Simple Mail Transfer Protocol (SMTP) Message Transfer Agent (MTA) has some native ability to control communications with other MTAs over the Internet. For example, sendmail, the most widely used SMTP MTA in the world, has the ability to disallow relaying and to limit the networks from which it accepts messages.

These features, referred to below as precursors and first-generation technologies, were not created specifically to address the spam problem, nor do they constitute an anti-spam technology. In the past few years, messaging product vendors seeking to address the spam problem have implemented rudimentary capabilities like these and crude features such as simple "white lists" and "black lists" into basic anti-spam feature sets. The latter approach, which is still evident in many products that include SMTP MTAs such as IBM Lotus Domino, does not truly represent an anti-spam technology.

True anti-spam technologies, referred to below as second and third generation technologies, are a relatively new development that followed the advent of spam in the mid 1990s. Anti-spam technologies are entirely new and are sharply differentiated from basic MTA controls. In particular, Bayesian algorithms and more recent advances in text analysis, for example, using Artificial Intelligence, are a radical departure from pre-existing mechanisms characterized or repurposed as anti-spam tools.

A short description of the successive technologies that have used to combat spam would cover the following generations.