arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.09326cs.CLcs.AIcs.SE
收藏

大语言模型抵御越狱攻击可靠性的表征与评估

Characterizing and Evaluating the Reliability of LLMs against Jailbreak Attacks

Zhejiang University(浙江大学);Nanyang Technological University(南洋理工大学)

Kexin Chen, Yi Liu, Dongxia Wang, Jiaying Chen, Wenhai Wang

AI总结:

本研究针对大语言模型的越狱攻击风险,构建多维度评估框架,对13款主流大语言模型开展大规模实证测试,给出可靠性评分并提出防护建议,为提升LLM安全评估提供参考。

摘要:

大语言模型(LLMs)在内容生成领域的作用日益关键,产生了显著的社会影响。这类模型存在生成可被认定为有害内容的潜在风险。为缓解这一风险,研究人员采取了包括设置安全防护机制在内的多项措施,确保大语言模型符合社会伦理规范。然而,即便有这些防护手段,“越狱”现象——即通过精心设计的提示词诱导模型生成有害回复——仍是一项严峻挑战。鉴于越狱手段持续带来的威胁及其对大语言模型可信使用的负面影响,严格评估模型抵御此类攻击的鲁棒性至关重要。\n本研究提出了一套综合评估框架,并开展了大规模实证实验以满足这一需求。我们聚焦于三大类共10种前沿越狱策略、来自61个特定有害类别的1525个问题,以及13款主流大语言模型。我们采用攻击成功率(ASR)、毒性评分、流畅度、Token长度、语法错误数等多维度指标,全面评估越狱攻击下大语言模型的输出结果。通过对这些指标进行归一化与聚合计算,我们给出了不同大语言模型的详细可靠性评分,并提出了降低模型此类漏洞易感性的策略建议。此外,我们还探究了模型、攻击策略、有害内容类型之间的关联,以及各评估指标之间的相关性,验证了我们多维度评估框架的有效性。\n我们的大量实验结果表明,所有受测大语言模型在应对特定策略时均缺乏抵御能力,凸显了关注大语言模型可靠性维度的必要性。我们相信本研究能为该领域内提升大语言模型抵御越狱攻击的安全评估工作提供有价值的参考。

英文摘要:

Large Language Models (LLMs) have increasingly become pivotal in content generation with notable societal impact. These models hold the potential to generate content that could be deemed harmful.Efforts to mitigate this risk include implementing safeguards to ensure LLMs adhere to social ethics.However, despite such measures, the phenomenon of "jailbreaking" -- where carefully crafted prompts elicit harmful responses from models -- persists as a significant challenge. Recognizing the continuous threat posed by jailbreaking tactics and their repercussions for the trustworthy use of LLMs, a rigorous assessment of the models' robustness against such attacks is essential. This study introduces an comprehensive evaluation framework and conducts an large-scale empirical experiment to address this need. We concentrate on 10 cutting-edge jailbreak strategies across three categories, 1525 questions from 61 specific harmful categories, and 13 popular LLMs. We adopt multi-dimensional metrics such as Attack Success Rate (ASR), Toxicity Score, Fluency, Token Length, and Grammatical Errors to thoroughly assess the LLMs' outputs under jailbreak. By normalizing and aggregating these metrics, we present a detailed reliability score for different LLMs, coupled with strategic recommendations to reduce their susceptibility to such vulnerabilities. Additionally, we explore the relationships among the models, attack strategies, and types of harmful content, as well as the correlations between the evaluation metrics, which proves the validity of our multifaceted evaluation framework. Our extensive experimental results demonstrate a lack of resilience among all tested LLMs against certain strategies, and highlight the need to concentrate on the reliability facets of LLMs. We believe our study can provide valuable insights into enhancing the security evaluation of LLMs against jailbreak within the domain.

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。

↑