The integration of Artificial Intelligence (AI) into the Internet of Things (IoT) ecosystem has transformed the landscape of cybersecurity. While IoT systems enable ubiquitous connectivity and automation across industries, they are also highly susceptible to cyberattacks due to their heterogeneity, limited resources, and scalability challenges. AI-based techniques, including machine learning, deep learning, and reinforcement learning, offer promising approaches to detecting anomalies, preventing intrusions, and predicting emerging threats in IoT networks. However, these opportunities are accompanied by significant challenges such as adversarial attacks, data privacy concerns, computational limitations, and the interpretability of AI models. This review article critically analyzes the dual role of AI in IoT security, highlighting its potential as both a defender and an enabler of cyber threats. Various AI-driven techniques are systematically reviewed, their applications in IoT security are discussed, and emerging risks are evaluated. The article further identifies future directions, emphasizing the importance of explainable AI, lightweight security frameworks, and robust adversarial defense mechanisms for sustainable and resilient IoT ecosystems.
Qadir et al. (Sat,) studied this question.