Round Corner
Department of Computer and Information Science


AI-based security attacks

AI and ML are widely used for pattern recognition, intrusion detection, malware classification, and therefore offer promising solutions in cyber-defense. However, recent study has shown that an attacker can leverage the neural processing to stealthily distribute an attack, by concealing malicious files into neural network model. Such attacks can evade existing security detection mechanisms and augment attacking capabilities. The focus of the thesis will include three parts:

- Literature review of malware insertion methods
- Create a vulnerable target system
- Studying how to use AI to create attacks on the target and how to defend against the attacks. For instance, attacks such as DeepLocker [1] that show the malicious use of AI in attack process

[1] Dhilung Kirat, Jiyong Jang, Marc Ph. Stoecklin, “DeepLocker - Concealing Targeted Attacks with AI Locksmithing”, Black Hat USA Conference 2018


Jingyue Li Jingyue Li
Associate Professor
105 IT-bygget
735 94484 
NTNU logo