Capture-the-Phish

Abstract—Phishing attacks remain one of the most prevalent and dangerous forms of social engineering in cybersecurity. They exploit human error to bypass technological defenses, compromising credentials and sensitive data. This paper presents Capture the Phish, an integrated AI-powered framework that combines transformer-based natural language processing, explainable AI (XAI), and privacy-preserving federated learning. The system transforms phishing detection into an adaptive, interactive learning environment that simultaneously enhances machine intelligence and human awareness. Through deep semantic modeling, contextual explanations, and decentralized model training, the framework provides a scalable, privacy-conscious, and explainable phishing defense system. Keywords—Phishing Detection, Explainable AI, Federated Learning, Cybersecurity, BERT, RoBERTa, Data Privacy.

Paper

Full text

PDF

Capture-the-Phish

Semantic Scholar · 2026

Abstract

Abstract—Phishing attacks remain one of the most prevalent and dangerous forms of social engineering in cybersecurity. They exploit human error to bypass technological defenses, compromising credentials and sensitive data. This paper presents Capture the Phish, an integrated AI-powered framework that combines transformer-based natural language processing, explainable AI (XAI), and privacy-preserving federated learning. The system transforms phishing detection into an adaptive, interactive learning environment that simultaneously enhances machine intelligence and human awareness. Through deep semantic modeling, contextual explanations, and decentralized model training, the framework provides a scalable, privacy-conscious, and explainable phishing defense system.

Keywords—Phishing Detection, Explainable AI, Federated Learning, Cybersecurity, BERT, RoBERTa, Data Privacy.

Similar papers

© 2026 NYSGPT2525 LLC