Chatbot for Phishing Detection Using NLP and Deep Learning

Abstract

Phishing remains a persistent and growing cybersecurity threat, exploiting human trust through deceptive emails, messages, and websites. Reports show phishing at record levels—over one million attacks observed in a single quarter of 2022—accounting for 80% of security incidents, causing approximately $17,700 in losses per minute as of 2023. Traditional anti-phishing measures struggle to address these sophisticated threats, as cybercriminals continuously adopt new tactics. This project presents an intelligent chatbot leveraging Natural Language Processing (NLP) and deep learning to detect phishing attempts in real-time. The chatbot, built with OpenAI’s GPT-3.5 language model, analyzes input text or images via Optical Character Recognition (OCR) to determine malicious intent. It provides users with clear explanations of its findings, enhancing both security and user awareness. Preliminary results demonstrate high accuracy in identifying phishing content, highlighting its significant potential to protect users effectively against scams. Key Terms – Chatbot, Deep Learning, Natural Language Processing, Phishing Detection.