arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2406.02619cs.CRcs.LG
收藏

通过密码学 Transformer 电路在语言模型中植入不可触发的后门

Unelicitable Backdoors in Language Models via Cryptographic Transformer Circuits

Contramont Research(康特拉蒙特研究院);IMCS UL(拉脱维亚大学数学与计算机科学学院(IMCS UL));Cavendish Labs(卡文迪什实验室);University of California, Berkeley(加利福尼亚大学伯克利分校);University of Oxford(牛津大学);Palisade Research(帕利塞德研究院)

Andis Draguns, Andrew Gritsevskiy, Sumeet Ramesh Motwani, Charlie Rogers-Smith, Jeffrey Ladish, Christian Schroeder de Witt

AI总结:

本文提出一种基于密码学技术的不可触发后门,使防御者无法在部署前触发或评估,且能抵御现有缓解策略,质疑了部署前检测的有效性。

摘要:

开源语言模型的快速普及显著增加了下游后门攻击的风险。这些后门可在模型部署期间引入危险行为,并能逃避传统网络安全监控系统的检测。在本文中,我们在 transformer 模型中引入了一类新型后门,与现有技术不同,其本质上具有不可触发(unelicitable)特性。不可触发性阻止防御者触发后门,使得即使在拥有完全白盒访问权限并使用自动化技术(如红队测试或某些形式化验证方法)的情况下,也无法在部署前进行正确评估。我们表明,得益于使用密码学技术,我们的新型构造不仅具有不可触发特性,而且具有良好的鲁棒性属性。我们在实证研究中证实了这些属性,并提供证据表明我们的后门能够抵御最先进的缓解策略。此外,我们在先前工作的基础上进行扩展,表明我们的通用后门虽然在白盒设置中并非完全无法检测,但比一些现有设计更难被检测到。通过证明将后门无缝集成到 transformer 模型中的可行性,本文从根本上质疑了部署前检测策略的有效性。这为 AI 安全与安保领域的攻防平衡提供了新见解。

英文摘要:

The rapid proliferation of open-source language models significantly increases the risks of downstream backdoor attacks. These backdoors can introduce dangerous behaviours during model deployment and can evade detection by conventional cybersecurity monitoring systems. In this paper, we introduce a novel class of backdoors in transformer models, that, in contrast to prior art, are unelicitable in nature. Unelicitability prevents the defender from triggering the backdoor, making it impossible to properly evaluate ahead of deployment even if given full white-box access and using automated techniques, such as red-teaming or certain formal verification methods. We show that our novel construction is not only unelicitable thanks to using cryptographic techniques, but also has favourable robustness properties. We confirm these properties in empirical investigations, and provide evidence that our backdoors can withstand state-of-the-art mitigation strategies. Additionally, we expand on previous work by showing that our universal backdoors, while not completely undetectable in white-box settings, can be harder to detect than some existing designs. By demonstrating the feasibility of seamlessly integrating backdoors into transformer models, this paper fundamentally questions the efficacy of pre-deployment detection strategies. This offers new insights into the offence-defence balance in AI safety and security.

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。

↑