ABSTRACT
Phishing is a fraudulent tactic where individuals or organizations impersonate legitimate entities to deceive internet users into divulging sensitive information. This exploit capitalizes on users’ technical naivety, effectively holding them hostage. Nowadays, attackers aim to fraudulently attack the personal information present in IoT devices through phishing links via e-mails, notifications, and messages. Devices like security cameras and wearable sensors are particularly susceptible, often falling victim to firmware-based attacks. The detection of phishing attacks may be accomplished using a variety of methods, such as visual similarity, heuristic, machine learning, deep learning, and lists-based strategies. In this chapter, the author proposed threat prediction models with different machine learning algorithms to identify phishing websites. The system uses the smote analysis technique in order to handle the imbalance numbers between phishing and legitimate classes. In terms of accuracy, precision, recall, and F1 scores for identifying phishing websites, the study reveals that the RM classifier outperforms others, achieving scores of 97.53%, 97.06%, 97.94%, and 97.50%, respectively. This proposed model serves as a valuable tool for IoT researchers, security engineers, and cyber threat policy-makers, enabling them to proactively protect IoT devices from the inherent risks associated with phishing attacks during the early stages of development.
