Developing a Real‑Time Dynamic Response System Against Phishing Attacks Based on Edge Computing and Predictive Intelligence

Abstract

This paper proposes a novel edge-based, predictive-intelligence system for real-time phishing attack detection and response. By integrating lightweight AI models with Edge Computing platforms and behavior monitoring, the system dynamically adapts to emerging threats. We evaluate its performance in terms of detection accuracy, latency, false-positive rates, and resource consumption, comparing it to traditional centralized models (e.g., GRU+WOA [1]). Results demonstrate superior detection speed (average latency 80 ms vs. 250 ms), higher accuracy (96.2% vs. 89.7%), and lower false positives (2.1% vs. 5.4%), while operating efficiently on edge devices.