Network security plays an important role in Internet of Things (IoT) security. The botnet attack is one of the challenges that has attracted the attention of experts in this domain in recent years. The (IoT)’s quick growth has changed the threat landscape, which raises the possibility of botnet attacks. In response, scientists using machine learning and deep learning enhanced the security of IoT networks. This comprehensive analysis investigates the existing status of machine learning approaches developed for detecting botnets in IoT scenarios. Specifically, this review clarifies methods, efficacy, and difficulties related to ML and DL techniques in thwarting IoT botnet invasions by synthesising recent scholarly contributions, including deep learning-based network traffic analysis, hybrid feature selection, and ensemble-based ML approaches. Additionally, it explores these publications investigating low-complexity models for detecting DDoS attacks and evaluating machine learning models using well-known datasets such as Bot-IoT, NBa-IoT, IOT-23 etc. As emphasised by the literature review, the dynamic nature of cyber threats and the Internet of Things highlights the urgent need for adaptive detection mechanisms. As a result, the review offers a snapshot of recent academic endeavours through a structured comparison table summarising key research attributes, facilitating knowledge dissemination, and directing future research in IoT botnet detection.