通过双射学习实现无尽越狱
Endless Jailbreaks with Bijection Learning
AI总结:
提出双射学习攻击,通过可控随机编码与上下文学习绕过 LLM 安全机制,并揭示模型能力越强、越易被高复杂度双射攻击严重越狱。
摘要:
尽管采取了广泛的安全措施,大语言模型(LLM)仍容易受到对抗性输入(即越狱攻击)的影响,这类输入可诱发不安全行为。在本研究中,我们提出双射学习(bijection learning),这是一种强大的攻击算法,能够利用随机生成且复杂度可被严格控制的编码,自动对 LLM 进行安全漏洞模糊测试。我们借助上下文学习向模型教授双射编码,将经过编码的查询传递给模型以绕过内置安全机制,最后再将响应解码回英文。我们的攻击在一系列前沿语言模型上都极为有效。此外,通过控制编码中键值映射数量等复杂度参数,我们发现受攻击 LLM 的能力水平与最有效双射攻击的平均复杂度之间存在密切关系。本研究强调,前沿模型中的新漏洞可能会随规模扩大而出现:能力越强的模型,被双射攻击越狱的严重程度越高。
英文摘要:
Despite extensive safety measures, LLMs are vulnerable to adversarial inputs, or jailbreaks, which can elicit unsafe behaviors. In this work, we introduce bijection learning, a powerful attack algorithm which automatically fuzzes LLMs for safety vulnerabilities using randomly-generated encodings whose complexity can be tightly controlled. We leverage in-context learning to teach models bijective encodings, pass encoded queries to the model to bypass built-in safety mechanisms, and finally decode responses back into English. Our attack is extremely effective on a wide range of frontier language models. Moreover, by controlling complexity parameters such as number of key-value mappings in the encodings, we find a close relationship between the capability level of the attacked LLM and the average complexity of the most effective bijection attacks. Our work highlights that new vulnerabilities in frontier models can emerge with scale: more capable models are more severely jailbroken by bijection attacks.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。