Use of Agentic AI with OpenAI and Prompt Engineering and State-of-the Art Machine Learning Algorithm to detect the patterns in IOT Device Network Intrusion Attacks
Chandrani Mukherjee · 2025
The Internet of things (IoT) is a system of interrelated computing devices including sensors, cameras, smart doors and many such mechanical and digital electronic devices having unique identifiers (UIDs) and could transfer data over a network without requiring human intervention. The Internet of Things (IoT) encompasses a network of interconnected computing devices such as sensors, cameras, smart doors, and various other mechanical and digital electronic devices, each uniquely identified and capable of transmitting data over a network autonomously. During pandemics like COVID-19, minimizing contact with objects in public spaces and gatherings is crucial, thus heightening the demand for IoT devices and sensors. These IoT devices, however, are susceptible to various cyber intrusions, necessitating robust Intrusion Detection Systems (IDS). Attackers can compromise IoT devices, creating botnets that accumulate valuable data and infiltrate systems. IoT devices face threats from malware such as Mirai, Trojan, Gafgyt, Okiru, and Hakai. This proposal focuses on device categories like door locks, Philips Hue smart bulbs, and Amazon Echo smart speakers. Previous research has explored IDS models for IoT using deep neural networks, classification techniques, explainable AI, and PCA, but none have leveraged Generative AI (GenAI) for such purposes. This study aims to demonstrate the application of Large Language Models (LLM) combined with Agentic AI to classify, explain, and implement actions in IDS for IoT. Generative AI, including ChatGPT, represents a powerful approach to addressing IDS attacks, enabling minimal model code in end devices while enhancing security measures.