The Next Frontier of IoT Security: Real-Time Phishing Defense Powered by Edge AI

Abstract

The rapid and substantial proliferation of Internet of Things (IoT) devices has facilitated the integration of artificial intelligence (AI) with edge computing, enabling intelligent, decentralized data processing to enable various applications. This study critically evaluates the advancements in AI-driven edge computing inside IoT ecosystems from 2020 to 2025, concentrating on phishing detection and mitigation. We examine peer-reviewed literature to assess not only cutting-edge AI technologies, including deep learning, federated learning, and reinforcement learning, but also architectural innovations pertinent to smart cities, healthcare, and industrial IoT. The advancements provide real-time data mining, high scalability, and energy-efficient system functionality, significantly enhancing the performance of IoT systems. It is noteworthy that edge AI can function as a facilitator for an effective phishing detector, enabling the identification and mitigation of threats locally and promptly, which is crucial for maintaining a secure IoT network in highly sensitive environments. Nevertheless, concerns such as security vulnerabilities, interoperability issues, poor latency, and limited resources characterize edge devices. The evaluation highlights the lack of standardized criteria for AI model implementation and insufficient defensive measures against sophisticated phishing attacks. The gaps hinder seamless integration and scalability in heterogeneous IoT systems. We propose future research avenues focused on developing adaptive AI models for a dynamic threat landscape, establishing standardized interoperability, and creating lightweight cryptographic solutions tailored for resource-constrained devices. The paper provides a comprehensive synthesis of existing situations, opportunities, and difficulties, offering research and practical insights to enhance the security and efficiency of AI-based edge computing in IoT applications. By addressing these problems, it is feasible to harness the complete potential of AI at the edge and transform IoT ecosystems into intelligent, resilient cyber networks capable of mitigating emerging cyber threats, including phishing.