Deep Learning-Based Intrusion Detection Systems for Phishing Email Detection: A Short Survey

Axel De Nardin, Silvia Zottin, Claudio Piciarelli, Gian Luca Foresti; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, 2025, pp. 7626-7634

Phishing remains a prevalent and evolving threat within the cybersecurity landscape, exploiting human vulnerabilities through deceptive email content. This survey presents a focused review of deep learning-based intrusion detection systems (IDSs) tailored to phishing email detection. It emphasizes recent innovations in neural architectures, including CNNs, RNNs, Transformer-based models, and hybrid or multi-modal systems, highlighting their design principles and comparative performance. We analyze a wide range of public and private phishing-related email datasets, assessing their scope, representativeness, and limitations in supporting generalizable detection models. Furthermore, we examine how these models cope with real-world deployment challenges, including adversarial manipulation, data imbalance, and the integration of multi-modal cues like URLs and headers. This work aims to guide future research by identifying critical gaps in robustness, scalability, and dataset diversity.