Offensive AI is the use of artificial intelligence for malicious purposes including attacks on AI systems (a.k.a. adversarial machine learning) and attacks powered by AI (deepfakes, swarm malware, ML for zero-day detection, and so on).
The Offensive AI Lab (OAI Lab) studies how adversaries use AI to enhance their attacks. The objective of the lab is to protect society by identifying and mitigating AI threats before they become mainstream. We are located in the BGU Cyber Security Research Center in Be’er Sheva Israel.
Featured Project
GAVEL brings rule-based, signature-style cyber defense to AI safety — auditable rules that monitor an LLM’s internal activations and block unsafe behavior without retraining the model. Accepted to ICLR 2026 (Rank A*) and presented at Black Hat USA 2026. Read more →
Our Supporters
This lab has is supported by the Zuckerman STEM Leadership Program
We would also like to thank the various organizations who have supported our research by funding projects in our lab:
