The Impact of Artificial Intelligence on Security: a Dual Perspective

This paper analyses the impact of Artificial Intelligence (AI) on security processes. Through the analysis of risk maps (a risk analysis tool), we highlight two opposing views: Beneficial AI and Malicious AI. Beneficial AI focuses on improving security, covering capabilities such as security design and testing assistance, system security monitoring, and decision making upon cyber-attacks. Malicious AI focuses on lowering security, covering capabilities such assistance for attack undetectability, or for attack decision making. While we recall means of attacks ranging from enhanced cyber-attacks to social-engineering, we also describe ways of integrating AI in companies and products’ life cycle and reflections about ethics in AI. We then analyze how impacted IoT systems may be considering the relationships between connected objects and AI models and their use cases. Finally, we conclude with two recommendations: revisiting risk frameworks to integrate AI, and providing recommendations for an ethical approach to AI research.