Adversarial Machine Learning in Cybersecurity
Vikram Sıngh, Sanyogita Singh · 2025
With the evolution and penetration of AI and ML into almost all critical public life domains including cybersecurity, the cybercrime ecosystem attempts to tap the vulnerabilities in AI-based cybersecurity systems by invoking adversarial machine learning, which has posed a significant challenge to cyber-physical security systems employing ML. Kitty of AML techniques, as the nomenclature itself indicates, take advantage of vulnerabilities in machine learning models. Cybercriminals manipulate input vectors to carry out their evil intent and evade their detection. The proposed chapter investigates the core concepts in adversarial machine learning, its implications in the arena of cybersecurity, and mitigation of these cyberthreats. Attack methodologies, detection strategies, and defence mechanisms have been studied in this chapter. Additionally, the ethical and legal challenges are discussed to comprehend the evolutionary discipline of adversarial machine learning in cyber-physical cyber-digital security.