arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.04459cs.CRcs.AIcs.SE
收藏

基准的基准:解构影响与代码仓库质量在LLM安全基准中的作用

Benchmark of Benchmarks: Unpacking Influence and Code Repository Quality in LLM Safety Benchmarks

CISPA Helmholtz Center for Information Security(CISPA海德堡信息安全研究中心);University of Waterloo(滑铁卢大学);Flexera(Flexera公司)

Junjie Chu, Xinyue Shen, Ye Leng, Michael Backes, Yun Shen, Yang Zhang

AI总结:

本文通过系统研究31个LLM安全基准及382篇非基准论文,发现仅39%的基准仓库可直接运行,且缺乏伦理考量。研究揭示社区采用基准与作者声望和代码运行性相关,但与代码质量无关,指出安全基准可能存在安全隐患和不可比性问题。

摘要:

LLM安全研究的快速扩展带来了跟踪进展的挑战,使基准成为识别关键趋势和系统比较的重要评估基础设施。然而,目前尚无对基准代码质量和运行性进行系统评估,也缺乏对社区为何选择某些基准而非其他基准的因素分析。为此,我们系统研究了31个LLM安全基准(涵盖提示注入、劫持和幻觉)及382篇非基准论文,结合自动化静态分析、人工运行性测试(220多人小时)和文献计量分析。研究发现,仅39%的基准仓库无需修改即可运行,仅16%提供无缺陷的安装指南,而6%的基准尽管包含潜在有害内容,仍缺乏伦理考量。这些缺陷在研究期间持续存在,无显著改善。分析采用因素时,发现基准采用与作者声望和代码运行性相关,但与代码质量标准如Pylint得分和可维护性无关,表明社区在选择基准时并未奖励更高的编码标准。基于这些结果,我们识别出潜在的安全和可靠性问题。一些安全基准仓库公开暴露有害内容,如成功的劫持响应,未有任何伦理警告或访问控制,实际上充当了无防护的攻击资源。此外,当基准需要临时修改才能运行时,不同论文中的下游安全评估可能不可比。本文通过案例研究说明这些具体后果,并提出有针对性的检查表,帮助基准贡献者改进代码质量、文档和伦理实践。

英文摘要:

The rapid expansion of research in LLM safety presents challenges in tracking advancements, making benchmarks important evaluation infrastructures for identifying key trends and facilitating systematic comparisons. Yet no systematic assessment exists of their code quality and runnability, nor of what factors are associated with the community's adoption of certain benchmarks over others. To address this gap, we conduct a systematic measurement study of 31 LLM safety benchmarks (covering prompt injection, jailbreak, and hallucination) with 382 non-benchmark papers as a control group, combining automated static analysis, human runnability testing (220+ person-hours), and bibliometric analysis. We find that only 39\% of benchmark repositories can run without modification, only 16\% provide flawless installation guides, and a mere 6\% include ethical considerations despite containing potentially harmful content. These deficiencies persist across the study period with no significant improvement. Analyzing adoption factors, we find that benchmark adoption correlates with author prominence and code runnability, but not with code quality standards such as Pylint score and maintainability, suggesting that the community's benchmark selection does not reward higher coding standards. Based on these results, we identify potential safety and reliability concerns. Some safety benchmark repositories openly expose harmful content, such as successful jailbreak responses, without any ethical warning or access control, effectively serving as unguarded attack resources. Furthermore, when benchmarks require ad-hoc modifications to run, downstream safety evaluations across different papers may not be comparable. We present case studies illustrating these concrete consequences and propose a targeted checklist to help benchmark contributors improve code quality, documentation, and ethical practices.

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。

↑