Evasion, Poisoning, Extraction, and Inference - Tools to Defend and Evaluate

Explore the world of AI security in this 40-minute technical talk from the RSA Conference. Delve into the advanced tools used by adversaries for evasion, poisoning, extraction, and inference attacks against machine learning models. Learn about open-source tools that can reproduce these attacks and discover defensive strategies to evaluate and protect AI applications before deployment. Join IBM Research Staff Members Beat Buesser and Abigail Goldsteen as they provide valuable insights into safeguarding AI systems against malicious actors.

Evasion, Poisoning, Extraction, and Inference: Tools to Defend and Evaluate