arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2511.04538cs.CL
收藏

从模型到漏洞:迈向可执行的LLM生成漏洞报告

From Model to Breach: Towards Actionable LLM-Generated Vulnerabilities Reporting

IEM, HES-SO Valais-Wallis(IEM,HES-SO瓦莱-达沃斯);II, HES-SO Valais-Wallis(II,HES-SO瓦莱-达沃斯);Cyber-Defence Campus, armasuisse(网络安全防御校区,armasuisse)

Cyril Vallez, Alexander Sternfeld, Andrei Kucharavy, Ljiljana Dolamic

AI总结:

本文提出提示暴露度(PE)和模型暴露度(ME)指标,用于量化LLM生成漏洞的严重性与普遍性,并揭示当前开放权重编码模型在现实场景中仍存在安全性与功能性权衡问题。

摘要:

随着基于大语言模型(LLM)的编码助手在软件开发中的作用日益关键,它们所生成的漏洞在整个网络安全格局中的影响也变得愈发重要。尽管已经提出了一系列LLM代码安全基准以及改进生成代码安全性的方法,但这些工作对广泛使用的编码LLM产生了多大影响仍不清楚。在此,我们表明,即便是最新的开放权重模型,在现实使用场景下最早被报告的漏洞场景中仍然存在脆弱性,这表明安全性与功能性之间的权衡至今阻碍了漏洞的有效修补。为了帮助解决这一问题,我们引入了一种新的严重性度量指标,该指标反映LLM生成漏洞所带来的风险,综合考虑漏洞严重性、生成概率以及诱导生成易受攻击代码的提示表述——即提示暴露度(Prompt Exposure, PE)。为了推动对最严重和最普遍漏洞的缓解,我们使用PE定义了模型暴露度(Model Exposure, ME)分数,该分数表示模型生成漏洞的严重程度和普遍性。

英文摘要:

As the role of Large Language Models (LLM)-based coding assistants in software development becomes more critical, so does the role of the bugs they generate in the overall cybersecurity landscape. While a number of LLM code security benchmarks have been proposed alongside approaches to improve the security of generated code, it remains unclear to what extent they have impacted widely used coding LLMs. Here, we show that even the latest open-weight models are vulnerable in the earliest reported vulnerability scenarios in a realistic use setting, suggesting that the safety-functionality trade-off has until now prevented effective patching of vulnerabilities. To help address this issue, we introduce a new severity metric that reflects the risk posed by an LLM-generated vulnerability, accounting for vulnerability severity, generation chance, and the formulation of the prompt that induces vulnerable code generation - Prompt Exposure (PE). To encourage the mitigation of the most serious and prevalent vulnerabilities, we use PE to define the Model Exposure (ME) score, which indicates the severity and prevalence of vulnerabilities a model generates.

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。

↑