Adversarial machine learning: A new frontier in cyber attacks
Nazmul Gony, IMRANUL HOQUE Bhuiyan, Md. Mostafizur Rahman, Mostafijur Rahman, Shayma Sultanan · World Journal of Advanced Research and Reviews · 2022
AML represents an emerging critical issue in cybersecurity that creates severe difficulties for security systems that use AI. AI-based security systems face increasing threats because threat responders use adversarial techniques to exploit vulnerabilities in these systems, as organizations depend on AI more often for threat detection and response. This analysis studies how AML poses an escalating threat to contemporary cyberattacks while affecting the operation of AI security models. A comprehensive analysis of real-world security cases alongside present defense methods allows this paper to reveal the principal flaws that affect AI system security performance. The paper presents recommendations to strengthen AI model resistance against adversarial attacks and suggests potential research directions within this field.