Fraud detection software typically employs a combination of machine learning, pattern recognition, and real-time analytics to identify potential threats. The software scans incoming and outgoing emails for anomalies and cross-references them with known fraud patterns and databases of malicious IP addresses. When a potential threat is detected, the software can either alert the user or automatically take action, such as quarantining the suspicious email.