Design & Evaluation of a Multi-Modal Deepfake Detection Framework for Advanced Phishing Threats

Corr, Gavin (2025) Design & Evaluation of a Multi-Modal Deepfake Detection Framework for Advanced Phishing Threats. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (380kB) | Preview

Abstract

Traditional phishing defence strategies such as text and URL analysis, are poorly equipped to detect advanced media manipulations, such as deepfakes. Deepfake technology has made phishing attacks more sophisticated, creating a challenge for traditional detection methods. This study looks at a lightweight solution for advanced deepfake detection. The framework uses two machine learning models, a convolutional neural network (CNN) for audio deepfake detection, and a MobileNetV2 based transfer learning model for visual deepfake detection. Both models are embedded into a browser-based application deployed publicly on Hugging Face using a Gradio UI. This allows users to upload audio, image or video files for classification, confidence scoring and CSV logging. Evaluation results show that the audio model achieves a 97% accuracy rating with perfect recall scores for fake samples, which allows it to detect synthetic speech reliably. The visual model achieves 86% accuracy with high precision but lower recall on fake media, highlighting the difficulty of single-frame forgery detection and indicating potential for future work on fusion and robustness.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Mustafa, Raza Ul UNSPECIFIED
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence Q Science > QA Mathematics > Computer software > Computer Security T Technology > T Technology (General) > Information Technology > Computer software > Computer Security Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Cyber Security
Depositing User:	Ciara O'Brien
Date Deposited:	03 Sep 2026 10:45
Last Modified:	03 Sep 2026 10:45
URI:	https://norma.ncirl.ie/id/eprint/9795

Actions (login required)

View Item