Phishing Attacks in the Age of Artificial Intelligence: A Systematic Survey

Abstract

One of the most dangerous frauds from today's techno-psychological era is phishing that plays on the vulnerabilities of technology and human psychology through email, SMS, telephone or social media and other new channels. Phishing attacks have evolved significantly with the advent of Artificial Intelligence, changing their nature, scope, and stealthiness, and continuing the arms race with defenders. The results reveal a number of points that need to be addressed, such as the growing prevalence of detections, the continued vulnerabilities within the ML systems and the necessity for multi-layered defense mechanisms that combine automated AI answers with controlled, human monitoring. Critical gaps include explainable AI, non-email vector coverage, and adversarial robustness benchmarks.