外观
论文arXiv 2607.282263 个月前
基于世界模型的具身 AI 安全：威胁、防御与评估的生命周期
原文链接：Security of World-Model-Based Embodied AI: A Lifecycle of Threats, Defenses, and Evaluation（arXiv 2607.28226 · 发布日期：2026-07-30）
Security of World-Model-Based Embodied AI: A Lifecycle of Threats, Defenses, and Evaluation
该综述面向基于世界模型的具身 AI 安全问题，按生命周期梳理威胁、防御与评估协议。
核心要点
- 世界模型支持预测与规划，但会把数据、传感器、提示或反馈风险传导到物理动作。
- 综述沿数据构建、表征学习、状态 grounding、想象、轨迹评估、执行与长期适应梳理威胁。
- 将投毒、后门、对抗样本、传感器欺骗、提示注入等攻击映射到世界模型安全属性。
- 提出生命周期分类、评估协议，并组织来源、鲁棒 grounding、不确定性预测等防御方向。
英文摘要 (Abstract)
World models give embodied AI a predictive core: they compress observations into states, simulate action-conditioned futures, and enable planning beyond reactive control. This predictive layer, however, opens a new security boundary-compromise can propagate from data, sensors, prompts, or feedback into physical action. Rather than treating world models as an isolated component, this survey traces threats across their entire lifecycle-from data construction and representation learning, through state grounding and imagination, to trajectory evaluation, execution, and long-term adaptation via memory and tools. We show that familiar attack families: poisoning, backdoors, adversarial examples, sensor spoofing, prompt injection, trajectory manipulation, and supply-chain attacks take on distinct meanings when they corrupt world states, learned dynamics, affordance estimates, or safety costs. We also highlight a duality: world models can serve as runtime safety shields, yet when compromised or over-trusted they generate predictive safety illusions. The survey offers a lifecycle taxonomy, maps existing attacks to world-model security properties, outlines evaluation protocols for safety failures, and structures defenses across provenance, robust grounding, uncertainty-aware prediction, trajectory gating, feedback auditing, and deployment assurance.
作者：Fazhong Liu, Zhuoyan Chen, Haozhen Tan, Yan Meng, Guoxing Chen, Haojin Zhu
来源
本页事实对应的公开页面，以原文为准。