Demo

Infecting LLM Agents via Generalizable Adversarial Attack
Weichen Yu, Kai Hu, Tianyu Pang, Chao Du, Min Lin, and Matt Fredrikson.
Red Teaming GenAI Workshop @ NeurIPS 2024 · Oral · Paper

An animated illustration of one initial intervention spreading through messages and memory across eight LLM agents. Adapted from the paper’s six-round example; outcomes depend on the model and context. 44 seconds · English.