arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2507.03473cs.CLcs.AI
收藏

超越武器化:中低资源语言自身的 NLP 安全

Beyond Weaponization: NLP Security for Medium and Lower-Resourced Languages in Their Own Right

Heather Lent

AI总结:

本文针对中低资源语言的 NLP 安全问题,将对抗攻击扩展到 70 种语言,评估单语与多语模型安全性,发现单语模型参数过小且多语能力并不总能保证安全,为低资源语言社区的安全部署提供关键考量。

摘要:

尽管越来越多的证据表明多语言能力很容易被武器化以攻击语言模型(LMs),但 NLP 安全领域的研究仍然以英语为中心。在保护语言模型方面,NLP 领域“英语优先”的惯例与网络安全中的标准程序相冲突,后者要求从业者预期并准备应对最坏情况。为了缓解 NLP 安全中的最坏情况,研究人员必须愿意面对语言模型安全中最薄弱的环节:低资源语言。据此,本文考察了低资源和中等资源语言的语言模型安全性。我们将现有的对抗性攻击扩展到多达 70 种语言,以评估这些语言的单语和多语语言模型的安全性。通过分析,我们发现单语模型在参数总量上往往过小,难以确保可靠的安全性;同时,尽管多语言能力有所帮助,但并不总能保证安全性的提升。最终,这些发现为低资源语言社区更安全地部署语言模型提供了重要考量。

英文摘要:

Despite mounting evidence that multilinguality can be easily weaponized against language models (LMs), works across NLP Security remain overwhelmingly English-centric. In terms of securing LMs, the NLP norm of "English first" collides with standard procedure in cybersecurity, whereby practitioners are expected to anticipate and prepare for worst-case outcomes. To mitigate worst-case outcomes in NLP Security, researchers must be willing to engage with the weakest links in LM security: lower-resourced languages. Accordingly, this work examines the security of LMs for lower- and medium-resourced languages. We extend existing adversarial attacks for up to 70 languages to evaluate the security of monolingual and multilingual LMs for these languages. Through our analysis, we find that monolingual models are often too small in total number of parameters to ensure sound security, and that while multilinguality is helpful, it does not always guarantee improved security either. Ultimately, these findings highlight important considerations for more secure deployment of LMs, for communities of lower-resourced languages.

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。

↑