Compromising The AI Agent Ecosystem Via Its "Universal Connector"
Security researchers examine technical aspects of Compromising The AI Agent Ecosystem Via Its "Universal Connector".
Summary
Black Hat session on Compromising The AI Agent Ecosystem Via Its "Universal Connector".
Related talks
Black Hat·27 min
Kailiang Ying · Tongbo Luo · Xinyu Xing · Xuguang (Luke) Liu
Presenters detail vulnerability discovery and exploitation in Demystify AI Security Products With a Universal Pluggable XAI Translator.
IEEE Symposium on Security and Privacy·13 min
Generative Models
Presenters detail vulnerability discovery and exploitation in Learning Universal Adversarial Perturbations with Generative Models.
The Linux Foundation·37 min
David Aronchick
Demonstrates universal AI: Execute Your Models Where Your Data (And Users) Are and evaluates key implications for security practitioners.