Beyond the Black Box - Revealing Adversarial Neural Patterns in LLMs

Overview

Uncover how LLM jailbreaks function through neural introspection techniques that decode internal states, revealing previously unknown attack patterns and mechanics in large language models.

---

Subjects

Computer Science