Smart Phishing Blocker

Abstract

Phishing is a serious cybersecurity risk that uses phony websites to trick people into disclosing private information. Conventional detection techniques, such rule-based systems and blacklists, have trouble spotting novel or disguised phishing assaults. In order to improve accuracy, this research provides an AI-powered phishing detection system that combines Long Short-Term Memory (LSTM) and Artificial Neural Networks (ANN) with a sandbox-based behavioral analysis module. Real-time URL inspection is carried out by the hybrid ANN–LSTM model, which classifies websites prior to loading by learning both sequential dependencies and static lexical patterns. An LSTM classifier is used in a safe sandbox environment to assess malicious intent while also monitoring dynamic behaviors like pop-ups, redirects, and concealed data collecting. . The system, which is implemented as a lightweight Chrome plugin, offers an effective and scalable web security solution by enabling proactive, low-latency, and adaptive protection against zero-day phishing assaults.