Phishing attacks have become a serious cybersecurity threat by exploiting deceptive websites, visual interfaces, ma licious QR codes, and fraudulent SMS messages to steal sensitive user information. Existing phishing detection solutions primarily focus on a single attack vector, limiting their effectiveness against modern multi-modal attacks. This paper proposes an Explain able AI-Based Multi-Modal Detection Framework for detecting phishing, visual phishing, quishing, and smishing attacks. The framework integrates machine learning, deep learning, natural language processing, and computer vision techniques to analyze URLs, webpage screenshots, QR codes, and SMS messages. An XGBoost classifier achieves 99% accuracy for phishing URL detection, while a Random Forest classifier with TF-IDF features achieves 98.4% accuracy for smishing detection. Visual phishing detection is performed using the MobileNetV3 model, supported by Optical Character Recognition (OCR) for text extraction and Grad-CAM for visual explainability, achieving 72.2% accuracy. Experimental results demonstrate that the proposed framework accurately detects multiple phishing attack vectors in real time, providing an effective and explainable solution for enhancing cybersecurity.

Phishing Detection,Visual Phishing,Quishing,Smishing,Explainable Artificial Intelligence (XAI),Multi-Modal Detection,Machine Learning,Deep Learning,MobileNetV3,Grad-CAM,OCR