Multimodal Phishing Detection on Social Media: Evaluating Text-Image Fusion
In Plain Terms
This paper tests whether combining the text and images of social media posts helps detect phishing. Across 93 fusion designs, a text-only model (F1 = 86.96%) beat every model that also used images (best F1 = 81.87%). Images in phishing posts often add noise, so adding a modality helps only when it carries a real signal.
Artifacts
No artifacts listed yet.
Related Papers
Cite this
APA Redwing, K., Ampel, B. M., & Chen, H. (2026). Multimodal Phishing Detection on Social Media: Evaluating Text-Image Fusion. In 21st Annual Pre-ICIS Workshop on Information Security and Privacy (WISP).
BibTeX
@inproceedings{redwing2026multimodal,
title={Multimodal Phishing Detection on Social Media: Evaluating Text-Image Fusion},
author={Kayla Redwing and Benjamin M. Ampel and Hsinchun Chen},
booktitle={21st Annual Pre-ICIS Workshop on Information Security and Privacy (WISP)},
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.