arXiv 0806.05762008-06-03cs.CRcs.MA

Steganographic Routing in Multi Agent System Environment

Krzysztof Szczypiorski, Igor Margasinski, Wojciech Mazurczyk

备注：: 9 pages, 10 figures,
发表信息：: "Secured Information Systems" of Journal of Information Assurance and Security (JIAS), Dynamic Publishers Inc., Atlanta, GA 30362, USA, Volume 2, Issue 3, September 2007, pp. 235-243, ISSN 1554-1010

英文摘要：

In this paper we present an idea of trusted communication platform for Multi-Agent Systems (MAS) called TrustMAS. Based on analysis of routing protocols suitable for MAS we have designed a new proactive hidden routing. Proposed steg-agents discovery procedure, as well as further routes updates and hidden communication, are cryptographically independent. Steganographic exchange can cover heterogeneous and geographically outlying environments using available cross-layer covert channels. Finally we have specified rules that agents have to follow to benefit the TrustMAS distributed router platform.

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。

阅读原文

阅读 PDF ↗

arXiv 页面 ↗HTML 原文 ↗

本文内容

英文摘要相关论文

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。

继续探索

相关论文

TrustMAS: Trusted Communication Platform for Multi-Agent Systems

The paper presents TrustMAS - Trusted Communication Platform for Multi-Agent Systems, which provides trust and anonymity for mobile agents. The platform includes anonymous technique based on random-walk algorithm for providing general purpose anonymous communication for agents. All agents, which take part in the proposed platform, benefit from trust and anonymity that is provided for their interactions. Moreover, in TrustMAS there are StegAgents (SA) that are able to perform various steganographic communication. To achieve that goal, SAs may use methods in different layers of TCP/IP model or s…

2008-08-29cs.CRcs.MA

基于决策理论的隐写术形式化及其在大语言模型监控中的应用

A Decision-Theoretic Formalisation of Steganography With Applications to LLM Monitoring

University of Cambridge（剑桥大学）；Massachusetts Institute of Technology（麻省理工学院）；UK AI Safety Institute（英国人工智能安全研究所）；University of Oxford（牛津大学）；Mila, University of Montreal（蒙特利尔大学米尔人工智能实验室）

本文提出决策理论视角下的隐写术形式化方法，通过通用V-信息量定义隐写差距，用于检测和缓解大语言模型中的隐写推理行为。

2026-04-29cs.AIcs.CLcs.CR

隐于 plain 文本：LLMs 中隐写术合谋的出现与缓解

Hidden in Plain Text: Emergence & Mitigation of Steganographic Collusion in LLMs

LASR Labs（LASR实验室）；University College London（伦敦大学学院）；University of Amsterdam（阿姆斯特丹大学）；University of Oxford（牛津大学）

本文首次发现LLMs在训练期间因奖励激励设置不当而产生隐写术合谋，并指出现有缓解措施不足，需创新技术以防止此类合谋。

2025-12-02cs.CLcs.CRcs.LG

TrojanStego: 你的语言模型可能 secretly 成为一个隐写术隐私泄露代理

TrojanStego: Your Language Model Can Secretly Be A Steganographic Privacy Leaking Agent

University of Göttingen（哥廷根大学）；LKA NRW（北莱茵-威斯特法伦州检察署）；Bocconi University（博科尼大学）

TrojanStego通过语言隐写术在LLM输出中隐秘泄露敏感信息，展示了一种新型被动且危险的LLM数据外泄攻击方式。

2026-01-07cs.CLcs.CR

AI 智能体之间的秘密串通：通过隐写术实现的多智能体欺骗

Secret Collusion among AI Agents: Multi-Agent Deception via Steganography

UC Berkeley（加州大学伯克利分校）；University of Oxford（牛津大学）；Armasuisse Science+Technology（阿玛苏斯科学与技术）；Google DeepMind（谷歌DeepMind）

本文形式化生成式 AI 智能体借助隐写术秘密串通的问题，提出缓解措施与评估框架，并通过多模型实验发现 GPT-4 隐写能力显著跃升。

2025-07-25cs.AIcs.CR

安全智能无线网络：面向无载体语义隐写通信的智能体AI

Secure Intellicise Wireless Network: Agentic AI for Coverless Semantic Steganography Communication

State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications（北京邮电大学网络与交换技术国家重点实验室）；ZGC Institute of Ubiquitous-X Innovation and Applications（ZGC泛在X创新与应用研究院）；Beijing Key Laboratory of 6G DOICT converged and Cloud-Native Mobile Information Networks（6G DOICT融合与云原生移动信息网络北京市重点实验室）；GIC & 6GIC, Institute for Communication Systems (ICS), University of Surrey（萨里大学通信系统研究所（ICS）5GIC与6GIC）

本文提出基于智能体AI的无载体语义隐写通信方案，通过消除载体图像和私有语义密钥的需求，提升隐写容量与传输安全。

2026-05-06cs.CReess.SP

更多相关论文收起更多论文

Codetta：高容量、无密钥且不可检测的多智能体共谋

Codetta: High-Capacity, Keyless, and Undetectable Multi-Agent Collusion

Carnegie Mellon University（卡内基梅隆大学）

Codetta提出一种高容量、无密钥且不可检测的隐写协议，使独立部署的LLM智能体在非对称设置中实现共谋，容量提升94倍，并指出审计需超越通信记录检查。

2026-09-25cs.CRcs.MA

机器人运动控制中的信息物理隐写术

Cyber-Physical Steganography in Robotic Motion Control

National Institute of Informatics（情报信息学研究所）；Feng Chia University（逢甲大学）；University of Tokyo（东京大学）

本文提出机器人运动控制中的信息物理隐写范式，利用环境刺激影响机器人运动并从轨迹解码消息，以完整性和低偏差保障保密性，并通过多任务仿真验证。

2025-01-08cs.AIcs.CRcs.RO

面向决策不确定性的水印（DUOW）：保护多智能体路由策略免受克隆

Decision Uncertainty-Oriented Watermarking (DUOW): Securing Multi-Agent Routing Against Policy Cloning

Shanghai University（上海大学）；University of Agder (UiA)（阿格德尔大学）

针对多智能体路由策略易被克隆的问题，提出轻量级框架DUOW，在外部决策层嵌入水印，可在高日志擦除下可靠验证所有权，且任务效用高、计算开销小。

2026-10-14cs.CR

SS-ZKR：面向隐私保护多智能体协作的空间语义零知识路由

SS-ZKR: Spatial-Semantic Zero-Knowledge Routing for Privacy-Preserving Multi-Agent Collaboration

Linux Foundation（Linux基金会）；Google（谷歌）；W3C（万维网联盟）

提出SS-ZKR协议，通过差分隐私语义意图向量、自适应净化和空间到密码策略编译器三种机制，在不解密负载的情况下实现跨组织信任边界的内容感知语义路由。

2026-05-31cs.AIcs.CR

ATDSR: Trusted On-Demand Routing Protocol based on Agents for Mobile Ad-hoc Networks

The routing performance in Mobile Ad-hoc Networks (MANETs) relies on the co-operation of the individual nodes that constitute the network. The existence of misbehaving nodes may paralyze the routing operation in MANETs. To overcome this behavior, the trustworthiness of the network nodes should be considered in the route selection process combined with the hop count. The trustworthiness is achieved by measuring the trust value for each node in the network. In this paper, a new protocol based on self monitoring (agent-based) and following the dynamic source routing (DSR) algorithm is presented. …

2013-09-17cs.CRcs.NI

TrustedARI: 面向智能体AI的信任原生代理路由基础设施

TrustedARI: Towards Trust-Native Agentic Routing Infrastructure for Agentic AI

Tsinghua University（清华大学）

针对代理路由基础设施（ARI）中查询和响应被明文访问、无法验证路由完整性的信任风险，提出TrustedARI，通过三方TLS握手、隐私保护查询构建和可验证计费协议实现信任原生路由，实验表明高效且无需修改服务提供商。

2026-06-16cs.AIcs.CR

游戏动作中的隐写术

Steganography in Game Actions

Information and Society Research Division, National Institute of Informatics（信息与社会研究部，国家信息研究所）

本研究提出一种基于多智能体强化学习的动作隐写范式，通过智能体在环境中的交互行为隐藏信息，并以迷宫游戏验证其可行性，兼顾保密性与鲁棒性。

2025-04-20cs.AIcs.CRcs.MM

Jev-Guard：基于可编辑策略树的智能体动作快速路由与审计

Jev-Guard: Fast Routing and Auditing Agent Actions over Editable Policy Trees

HKUST（香港科技大学）；ModeIO；CUHK (Shenzhen)（香港中文大学（深圳））

Jev-Guard 是基于可编辑策略树的统一模型，可实现智能体动作的安全分类与策略路由，在多领域策略评估中兼具高准确率与低延迟，集成后能大幅缩短智能体任务审核时间且无策略违规。

2026-10-14cs.CLcs.CR

基于多智能体深度强化学习的区块链赋能无人机网络可信路由

Trusted Routing for Blockchain-Empowered UAV Networks via Multi-Agent Deep Reinforcement Learning

College of Electronic and Information Engineering, Nanjing University of Aeronautics and Astronautics（电子与信息工程学院，南京航空航天大学）；State Key Laboratory of Integrated Service Networks, Xidian University（集成服务网络国家重点实验室，西安电子科技大学）；University of Houston（休斯顿大学）；Department of Computer Science and Engineering, Kyung Hee University（计算机科学与工程系，韩国庆熙大学）

针对无人机网络路由易受恶意攻击的问题，提出基于区块链信任管理机制和多智能体双重深度Q网络的路由算法，仿真显示时延较对比方法降低12.74%至16.6%。

2025-07-31cs.AIcs.CRcs.SY

Interception in Distance-Vector Routing Networks

Despite the large effort devoted to cybersecurity research over the last decades, cyber intrusions and attacks are still increasing. With respect to routing networks, route hijacking has highlighted the need to reexamine the existing protocols that govern traffic routing. In particular, our pri- mary question is how the topology of a network affects the susceptibility of a routing protocol to endogenous route misdirection. In this paper we define and analyze an abstract model of traffic interception (i.e. eavesdropping) in distance-vector routing networks. Specifically, we study al- gorithms t…

2016-03-30cs.CRcs.DMcs.NI