Weaponized AI for cyber attacks

Abstract

Artificial intelligence (AI)-based technologies are actively used for purposes of cyber defense. With the passage of time and with decreasing complexity in implementing AI-based solutions, the usage of AI-based technologies for offensive purposes has begun to appear in the world. These attacks vary from tampering with medical images using adversarial machine learning for false identification of cancer to the generation of adversarial traffic signals for influencing the safety of autonomous vehicles. In this research, we investigated recent cyberattacks that utilize AI-based techniques and identified various mitigation strategies that are helpful in handling such attacks. Further, we identified existing methods and techniques that are used in executing AI-based cyberattacks and what probable future scenarios will be plausible to control such attacks by identifying existing trends in AI-based cyberattacks.