arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.10260cs.LGcs.CLcs.CR
收藏

针对无效越狱的 StrongREJECT

A StrongREJECT for Empty Jailbreaks

Center for Human-Compatible AI(人类兼容人工智能中心);UC Berkeley(加州大学伯克利分校)

Alexandra Souly, Qingyuan Lu, Dillon Bowen, Tu Trinh, Elvis Hsieh, Sana Pandey, Pieter Abbeel, Justin Svegliato, Scott Emmons, Olivia Watkins, Sam Toyer

AI总结:

针对越狱攻击有效性常被夸大的问题,本文提出了 StrongREJECT 基准,通过高质量数据集与自动化评估器准确衡量越狱性能,并发现绕过安全微调的越狱会降低模型能力。

摘要:

大多数越狱论文声称它们提出的越狱方法非常有效,经常吹嘘接近 100% 的攻击成功率。然而,越狱开发者大幅夸大其越狱方法有效性的情况或许比比皆是。我们建议,出现这个问题的原因是,越狱研究人员缺乏一个标准、高质量的基准来评估越狱性能,只能自行创建。为了创建基准,研究人员必须选择一个包含禁止提示的数据集,受害模型将响应这些提示,同时还需要一种评估方法来对受害模型响应的有害性进行评分。我们表明,现有基准存在显著的缺陷,并引入了 StrongREJECT 基准来解决这些问题。StrongREJECT 的数据集包含要求受害模型必须提供具体、有害信息的提示,而其自动化评估器则衡量响应向禁止提示提供有用信息的程度。在此过程中,StrongREJECT 评估器在评估越狱有效性方面实现了与人类判断的最先进一致性。值得注意的是,我们发现与人类判断和 StrongREJECT 评估器相比,现有的评估方法显著夸大了越狱的有效性。我们描述了一种令人惊讶的新现象来解释这种差异:绕过受害模型安全微调的越狱往往会降低其能力。总而言之,我们的研究结果强调,研究人员在开发新的越狱攻击时需要使用高质量的基准,例如 StrongREJECT。我们在 https://strong-reject.readthedocs.io/en/latest/ 上发布了 StrongREJECT 的代码和数据。

英文摘要:

Most jailbreak papers claim the jailbreaks they propose are highly effective, often boasting near-100% attack success rates. However, it is perhaps more common than not for jailbreak developers to substantially exaggerate the effectiveness of their jailbreaks. We suggest this problem arises because jailbreak researchers lack a standard, high-quality benchmark for evaluating jailbreak performance, leaving researchers to create their own. To create a benchmark, researchers must choose a dataset of forbidden prompts to which a victim model will respond, along with an evaluation method that scores the harmfulness of the victim model's responses. We show that existing benchmarks suffer from significant shortcomings and introduce the StrongREJECT benchmark to address these issues. StrongREJECT's dataset contains prompts that victim models must answer with specific, harmful information, while its automated evaluator measures the extent to which a response gives useful information to forbidden prompts. In doing so, the StrongREJECT evaluator achieves state-of-the-art agreement with human judgments of jailbreak effectiveness. Notably, we find that existing evaluation methods significantly overstate jailbreak effectiveness compared to human judgments and the StrongREJECT evaluator. We describe a surprising and novel phenomenon that explains this discrepancy: jailbreaks bypassing a victim model's safety fine-tuning tend to reduce its capabilities. Together, our findings underscore the need for researchers to use a high-quality benchmark, such as StrongREJECT, when developing new jailbreak attacks. We release the StrongREJECT code and data at https://strong-reject.readthedocs.io/en/latest/.

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。

↑