Adaptive Phishing URL Classification: A Generative Adversarial Approach
In Plain Terms
Phishing detectors trained on old data quietly lose their edge as attackers change tactics. This paper trains a generator to invent realistic new phishing URLs and pits it against a detector, so the detector learns to handle attacks it has never seen. Tested against a private feed of current phishing links from 2024–2025, the adversarially trained detector caught about 5% more modern phishing URLs, showing this cat-and-mouse training helps classifiers keep up with evolving threats.
Artifacts
Related Papers
Cite this
APA Abdellatif, N., Wagner, M., Ampel, B. M., Hu, J., Ahmad-Post, Z., & Chen, H. (2026). Adaptive Phishing URL Classification: A Generative Adversarial Approach. In Proceedings of the 2026 IEEE Cyber Awareness and Research Symposium (CARS).
BibTeX
@inproceedings{abdellatif2026adaptive,
title={Adaptive Phishing URL Classification: A Generative Adversarial Approach},
author={Noah Abdellatif and Mason Wagner and Benjamin M. Ampel and James Hu and Zara Ahmad-Post and Hsinchun Chen},
booktitle={Proceedings of the 2026 IEEE Cyber Awareness and Research Symposium (CARS)},
year={2026}
}
Benjamin M. Ampel
Assistant Professor in Computer Information Systems and Director, Center for CyberAI Research (CCAIR)
My research focuses on AI-enabled Cybersecurity, including Cyber Threat Intelligence, Large Language Models, and Phishing Detection.