Abstract
There is a huge potential for artificial intelligence (AI)-driven attacks on internet of things (IoT) networks especially with the growing use of generative AI (GenAI) and regenerative AI (RGenAI). GenAIs are known for generating new datasets with similar attributes to its underlying trained datasets – while RGenAIs could reshape and adapt the inputs feeding GenAIs – based on feedback responses. These AI technologies, when combined, can inflict massive costs on IoT networks which may be difficult to detect and resolve. In this paper, a contextual model is proposed to examine the possible scenarios of IoT network attacks arising from the combination of GenAI/RGenAI entities. This model constructs AI-driven attacks on IoT ecosystems as a looped process occurring in three phases: learn, deploy, and adapt. The paper further considers the likely use cases of this proposed model and provides insight into the possible countermeasures to avoid such AI-attacks.