Policy and Safety · Models
提出一种自进化多智能体框架防御LLM越狱攻击
论文提出一种自进化多智能体框架防御 LLM 越狱攻击，通过外部记忆和提示将攻击失败抽象为方法级规则，无需参数更新，适用于开源和黑盒模型。
阅读 arXiv 原文为什么重要
该框架为动态适应新型越狱攻击提供了轻量级解决方案，有助于提升大语言模型的安全性。
关键事实
事实 1
大多数现有防御是静态的，其安全行为在部署时固定。
来源与依据
yet most remain static: their safety behavior is fixed at deployment, so they cannot accumulate defensive experience or adapt to unseen strategies
事实 2
该框架在攻击成功后将失败抽象为方法级规则。
来源与依据
when an attack succeeds, the framework abstracts that failure into a method-level rule capturing the structural attack wrapper rather than the harmful topic
事实 3
该机制完全通过外部记忆和提示操作，不更新参数。
来源与依据
The mechanism operates entirely through external memory and prompting, with no parameter updates, and applies to both open-weight and black-box API models