**What Intermediate Layers Know: Detecting Jailbreaks from Entropy Dynamics** >

**What Intermediate Layers Know: Detecting Jailbreaks from Entropy Dynamics**

> Jailbreak attacks reveal a persistent weakness in aligned Large Language Models: carefully crafted prompts can elicit policy-violating responses despite safety training. While most defenses operate at the prompt or output level, it remains unclear how harmful intent is encoded within the model's internal representations. We investigate this question by analyzing token-level predictive entropy trajectories across laye...

**来源信息**
- **来源**：HuggingFace Daily Papers（社区热门论文）
- **分类**：行业
- **发布时间**：2026-06-25 08:00（北京时间）
- **原文**：[打开原文](https://huggingface.co/papers/2606.25182)