August 26
Attacks on AI made easy
Track 2

The results produced by AI are more often than not astonishing, but nevertheless, machines are not flawless. The cost of mistakes increases when dealing with security—for hackers, AI can be just an additional high‑impact attack vector.

Thus, this talk is going to cover a few fundamental black‑box attacks on AI systems and the process of automating security testing for machine learning models using open‑source tools.