The Evolution of Phishing Attacks and Social Engineering Countermeasures in the AI Era: A Comprehensive Review

The emergence of Generative Artificial Intelligence (GenAI) and Large Language Models (LLMs) has fundamentally transformed the cyber threat landscape, shifting social engineering and phishing attacks from traditional patterns toward dynamic, sophisticated, and highly personalized scenarios. This paper adopts a comprehensive approach to review the evolution of phishing methodologies and their corresponding countermeasures in the AI era. To this end, a novel taxonomy of modern attacks—including advanced spear phishing, whaling, and AI-driven automation—is first presented. Subsequently, key attacker enablers, such as audio-visual deepfakes, jailbreaking of language models, and automated Open-Source Intelligence (OSINT) gathering processes, are analyzed. On the defensive front, this study examines intelligent countermeasures based on Machine Learning (ML), Natural Language Processing (NLP), and advanced real-time behavioral analytics. Furthermore, adversarial challenges, evasion of email filters, and the necessity of transforming next-generation security awareness training are thoroughly explored. Finally, by outlining future trends, the critical role of integrating AI with Zero Trust Architecture to proactively mitigate threats is established as a strategic imperative. The findings of this review demonstrate that overcoming these emerging threats requires a transition from static tools toward multi-layered, dynamic, and adaptive defense frameworks.

کلیدواژه ها:

Modern Phishing ، Generative AI ، Social Engineering ، Zero Trust Architecture ، Machine Learning ، Deepfake