In this paper the development of anti-spam software detecting information attacks is offered. For this purpose it is considered spam filtration system with the multilayered, multivalent architecture, coordinating all ISP's in the country. All users and ISPs of this system involved in spam filtration process. After spam filtering process, saved spam templates are analyzed and classified. This parameterizing of spam templates give possibility to define the thematic dependence from geographical. For example, what subjects prevail in spam messages sent from the certain countries? Analyzing origins of spam templates from spam-base, it is possible to define and solve the organized social networks of spammers.