NeurIPS 2026 口头报告论文提出智能体安全是系统问题
Agent Security is a Systems Problem
一组网络安全研究者在 arXiv 论文（arXiv:2605.18991，已被 NeurIPS 2026 接收为口头报告）中提出，智能体安全应作为系统问题处理，把驱动智能体的 AI 模型视为不可信组件，在系统层面强制执行安全不变量。论文基于操作系统、网络、形式化方法和对抗机器学习的经验总结出核心设计原则，分析了 11 个代表性真实攻击并说明系统原则如何阻止它们，同时指出落地这些原则面临的研究挑战。
Authors:Mihai Christodorescu, Earlence Fernandes, Ashish Hooda, Somesh Jha, Johann Rehberger, Kamalika Chaudhuri, Xiaohan Fu, Khawaja Shams, Guy Amir, Jihye Choi, Sarthak Choudhary, Nils Palumbo, Andrey Labunets, Nishit V. Pandya
Abstract:We take the position that agent security must be approached as a systems problem: the AI model powering the agent must be treated as an untrusted component, and security invariants must be enforced at the system level. Through this lens, efforts to increase model robustness (the dominant viewpoint in the community) are insufficient on their own. Instead, we must complement existing efforts with techniques from the systems security domain. Based on our experience as cybersecurity researchers in operating systems, networks, formal methods, and adversarial machine learning, we articulate a set of core principles, grounded in decades of systems security research, that provide a foundation for designing agentic systems with predictable guarantees. As evidence, we analyze eleven representative real-world attacks on agents and discuss how systems principles, if realized, could have prevented these attacks. We also identify the research challenges that stand in the way of implementing these principles in agents.
| Comments: | Accepted to NeurIPS 2026 (Oral Presentation) |
| Subjects: | Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2605.18991 [cs.CR] |
| (or arXiv:2605.18991v3 [cs.CR] for this version) | |
| https://doi.org/10.48550/arXiv.2605.18991 arXiv-issued DOI via DataCite |
Submission history
From: Nils Palumbo [view email]
[v1]
Mon, 18 May 2026 18:11:17 UTC (9,744 KB)
[v2]
Wed, 20 May 2026 17:25:38 UTC (9,744 KB)
[v3]
Wed, 4 Nov 2026 00:02:12 UTC (9,749 KB)
来源：arXiv cs.AI · arxiv.org