arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.01414cs.RO
收藏

通过动作层面操控实现体感大语言模型的突破

Jailbreaking Embodied LLMs via Action-level Manipulation

The Hong Kong Polytechnic University, Hong Kong SAR, China(香港理工大学);University of Cambridge, Cambridge, United Kingdom(剑桥大学)

Xinyu Huang, Qiang Yang, Leming Shen, Zijing Ma, Yuanqing Zheng

AI总结:

Blindfold通过动作层面操控实现对体感大语言模型的突破,其攻击成功率显著高于现有基线,凸显了对后果感知防御机制的迫切需求。

摘要:

体感大语言模型(LLMs)通过自然语言指令和动作与物理世界互动。然而,除了LLMs本身固有的语言层面风险,具有真实世界执行能力的体感LLMs引入了新的漏洞:看似语义无害的指令仍可能导致危险的现实后果,揭示了语言安全与物理结果之间的根本不匹配。在本文中,我们引入了Blindfold,一个自动攻击框架,利用体感LLMs在现实动作上下文中有限的因果推理能力。与其通过迭代试错对黑盒体感LLMs进行突破,Blindfold采用对抗性代理规划策略:它妥协一个本地代理LLM执行看似语义安全但执行时可能产生有害物理效果的动作操控。Blindfold进一步通过注入精心设计的噪声来隐藏关键恶意动作以逃避防御机制的检测,并结合基于规则的验证器来提高攻击可行性。在体感AI模拟器和真实世界6DoF机械臂上的评估表明,Blindfold的攻击成功率比现有最佳基线高53%,突显了需要超越表面语言审查,向后果感知的防御机制发展以确保体感LLMs安全的紧迫性。

英文摘要:

Embodied Large Language Models (LLMs) enable AI agents to interact with the physical world through natural language instructions and actions. However, beyond the language-level risks inherent to LLMs themselves, embodied LLMs with real-world actuation introduce a new vulnerability: instructions that appear semantically benign may still lead to dangerous real-world consequences, revealing a fundamental misalignment between linguistic security and physical outcomes. In this paper, we introduce Blindfold, an automated attack framework that leverages the limited causal reasoning capabilities of embodied LLMs in real-world action contexts. Rather than iterative trial-and-error jailbreaking of black-box embodied LLMs, Blindfold adopts an Adversarial Proxy Planning strategy: it compromises a local surrogate LLM to perform action-level manipulations that appear semantically safe but could result in harmful physical effects when executed. Blindfold further conceals key malicious actions by injecting carefully crafted noise to evade detection by defense mechanisms, and it incorporates a rule-based verifier to improve the attack executability. Evaluations on both embodied AI simulators and a real-world 6DoF robotic arm show that Blindfold achieves up to 53% higher attack success rates than SOTA baselines, highlighting the urgent need to move beyond surface-level language censorship and toward consequence-aware defense mechanisms to secure embodied LLMs.

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。

↑