The Internet of Things (IoT) sensor devices generate vast amounts of data that can be utilized by advanced Machine Learning (ML) algorithms for threat detection. However, due to limited computational resources and weak protection capabilities, they are vulnerable to various cyberattacks. With the increase in IoT appliances and advancement in Generative Artificial Intelligence (GenAI), IoT networks are encountering highly refined attacks such as deepfakes, poisoning data, spoofing sensor readings, and bypassing authentication. Further, the sophisticated adversarial attacks can even compromise ML-based security models, leaving IoT networks increasingly vulnerable. Hence, there is a need for advanced defensive techniques against these progressive attacks. Nevertheless, GenAI itself plays supportive role in enhancing IoT security by addressing data scarcity through synthetic data generation. It also improves intrusion detection systems for effective threat detection by training models on synthetic samples of actual attack scenarios. Additionally, it helps in privacy-preserving techniques for safe data exchange in IoT networks. Since GenAI can be used both offensively and defensively, it has thus become a mixed blessing for IoT cybersecurity.