Artificial Intelligence methods:

• Linguistic heuristics, based on special term databases and mathematics.
• Signatures (linguistic "fingerprints") of known non-compliance messages employing comparison algorithms.

Rule-driven methods:

• A set of formal rules based on the analysis of message headers, size, sender, and other criteria.
• Usage of blacklists based on checking message sender IP and e-mail addresses against several real-time blacklists available from government agencies through the Internet.

All these detection methods are used simultaneously, resulting in a high detection rate and low degree of false alarms. Furthermore, all these methods are controlled from a single on-site administration point that enables fine-tuning and customization.