PyBot/resources/JSON/seebug.json

122 lines
10 KiB
JSON
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

[
{
"title": "GLiGuard面向大语言模型安全防护的模式条件分类方法",
"link": "https://paper.seebug.org/3485",
"description": "作者Urchade Zaratiana, Mary Newhauser, George Hurn-Maloney, Ash Lewis 译者知道创宇404实验室翻译组 原文链接https://arxiv.org/html/2605.07982v1 摘要 保障大语言模型LLM输出安全合规、符合政策要求需要能跨多安全维度实时扩展的内容审核机制。然而当前最优的安全护栏模型均基于70亿27...",
"pubDate": "Mon, 11 May 2026 16:40:13 +0800",
"guid": "https://paper.seebug.org/3485",
"category": "AI安全"
},
{
"title": "基于大语言模型的 AI 智能体安全威胁与防御系统性综述——分层攻击面框架",
"link": "https://paper.seebug.org/3484",
"description": "作者Kexin Chu 译者知道创宇404实验室翻译组 原文链接https://arxiv.org/pdf/2604.23338v2 摘要 智能体AI系统可跨多会话规划、保留记忆、调用外部工具并与对等智能体协同无状态大语言模型则不具备这些能力。现有安全分类体系按攻击类型如提示注入、越狱划分威胁却未说明脆弱的架构组件或威胁显现的时间尺度。本文直接解决这些结构性问题提出分层攻击面模型...",
"pubDate": "Sat, 09 May 2026 16:53:56 +0800",
"guid": "https://paper.seebug.org/3484",
"category": "AI安全"
},
{
"title": "ReTokSync面向生成式语言隐写的自同步分词消歧方法",
"link": "https://paper.seebug.org/3481",
"description": "作者Yaofei Wang, Rui Wang, Weilong Pang等 译者知道创宇404实验室翻译组 原文链接https://arxiv.org/html/2604.25486v1 摘要 生成式语言隐写GLS通过将秘密信息嵌入自然语言生成过程实现隐蔽通信。然而在实际部署中GLS易受分词歧义影响相同的表层文本在接收端可能被重新分词为不同的token序列破坏通信双方共享的解码状...",
"pubDate": "Fri, 08 May 2026 16:42:48 +0800",
"guid": "https://paper.seebug.org/3481",
"category": "AI安全"
},
{
"title": "素域 PINI后量子 NTT 掩码的机器验证合成定理",
"link": "https://paper.seebug.org/3480",
"description": "作者Ray IskanderKhaled Kirah 译者知道创宇404实验室翻译组 原文链接https://arxiv.org/abs/2604.25878v1 摘要 本文是后量子密码掩码数论变换NTT硬件形式化验证分析系列论文的第六篇。第一篇论文[1]建立了 QANARY 平台的结构依赖分析,第二篇论文[2]量化了部分 NTT 掩码下的安全裕度。布尔掩码的合成理论通过无干扰NI...",
"pubDate": "Wed, 06 May 2026 16:44:36 +0800",
"guid": "https://paper.seebug.org/3480",
"category": "AI安全"
},
{
"title": "Google DeepMindAI 智能体陷阱",
"link": "https://paper.seebug.org/3479",
"description": "作者Matija Franklin, Nenad Tomašev等 译者知道创宇404实验室翻译组 原文链接https://blog.qiaomu.ai/api/images/document/2026/04/603f0cae8cd8ab94-ai-agent-trap.pdf 摘要 自主AI智能体日益在网络中自主行动它们面临一种全新挑战信息环境本身。由此产生了一个关键安全漏洞我们称之...",
"pubDate": "Fri, 24 Apr 2026 16:54:27 +0800",
"guid": "https://paper.seebug.org/3479",
"category": "AI安全"
},
{
"title": "合法终端管理软件遭滥用:疑似银狐攻击事件分析与溯源",
"link": "https://paper.seebug.org/3478",
"description": "作者:知道创宇高级威胁情报团队 一、事件概述 近期,我们在客户现场应急响应中排查到一起新型攻击事件。攻击者伪造常用工具安装包诱导执行,随即部署一款带有合法数字签名的终端管理软件。经技术溯源确认,该程序具备主机信息收集、远程控制等完整恶意能力,其 C2 基础设施与 “银狐” 高度关联。由于合法数字签名的天然 “免杀” 特性,该恶意程序可轻易绕过主流杀毒软件检测,实现隐蔽入侵与长期控制。 银狐简介 银...",
"pubDate": "Wed, 15 Apr 2026 03:31:00 +0800",
"guid": "https://paper.seebug.org/3478",
"category": "威胁情报"
},
{
"title": "SkillTrojan针对基于技能的智能体系统的后门攻击",
"link": "https://paper.seebug.org/3477",
"description": "作者Yunhao Feng, Yifan Ding, Yingshui Tan等 译者知道创宇404实验室翻译组 原文链接https://arxiv.org/html/2604.06811v1/https://arxiv.org/html/2604.06811v1 摘要 基于技能的智能体系统通过组合可复用技能完成复杂任务,在提升模块化与可扩展性的同时,引入了尚未被充分研究的安全攻击面。本文提...",
"pubDate": "Thu, 09 Apr 2026 08:26:00 +0800",
"guid": "https://paper.seebug.org/3477",
"category": "AI安全"
},
{
"title": "GUARDSLM面向小语言模型、基于令牌激活的越狱攻击防御方法",
"link": "https://paper.seebug.org/3476",
"description": "作者Md. Jueal Mia1, Joaquin Molto1, Yanzhao Wu1, M. Hadi Amini 译者知道创宇404实验室翻译组 原文链接https://arxiv.org/html/2603.28817v1/https://arxiv.org/html/2603.28817v1 摘要 小语言模型SLM正成为大语言模型LLM高效且经济可行的替代方案在计算成本...",
"pubDate": "Fri, 03 Apr 2026 02:48:00 +0800",
"guid": "https://paper.seebug.org/3476",
"category": "AI安全"
},
{
"title": "静默颠覆:通过卫星系统供应链植入物实施的传感器欺骗攻击",
"link": "https://paper.seebug.org/3475",
"description": "作者Jack Vanlyssel, Gruia-Catalin Roman, Afsah Anwar 译者知道创宇404实验室翻译组 原文链接https://arxiv.org/html/2603.10388v1/https://arxiv.org/html/2603.10388v1 摘要 欺骗攻击是地面系统最具破坏性的网络威胁之一,而在太空中这类威胁愈发危险——卫星难以在轨维修,且运营方依...",
"pubDate": "Tue, 24 Mar 2026 09:47:00 +0800",
"guid": "https://paper.seebug.org/3475",
"category": "AI安全"
},
{
"title": "增强网络入侵检测系统:一种抵御对抗攻击的多层集成方法",
"link": "https://paper.seebug.org/3474",
"description": "作者Nasim Soltani, Shayan Nejadshamsi等 译者知道创宇404实验室翻译组 原文链接https://arxiv.org/html/2603.10413v1/https://arxiv.org/html/2603.10413v1 摘要 对抗样本会对机器学习ML算法构成严重威胁。若被用于操控基于机器学习的网络入侵检测系统NIDS行为将危及网络安全。本研究旨...",
"pubDate": "Mon, 23 Mar 2026 08:31:00 +0800",
"guid": "https://paper.seebug.org/3474",
"category": "AI安全"
},
{
"title": "CUDA Agent面向高性能 CUDA 内核生成的大规模智能体强化学习",
"link": "https://paper.seebug.org/3473",
"description": "作者Weinan Dai, Hanlin Wu, Qiying Yu等 译者知道创宇404实验室翻译组 原文链接https://arxiv.org/html/2602.24286v1/https://arxiv.org/html/2602.24286v1 摘要 GPU内核优化是现代深度学习的基础但仍是一项高度专业化的任务需要深厚的硬件专业知识。尽管大语言模型LLM在通用编程任务中表现...",
"pubDate": "Tue, 17 Mar 2026 09:50:00 +0800",
"guid": "https://paper.seebug.org/3473",
"category": "AI安全"
},
{
"title": "Unmasking SilverFoxs New Trends: Decoding Evasion Tactics, Domain Impersonation, and Mass-Generated Fake Software",
"link": "https://paper.seebug.org/3472",
"description": "Author: Knownsec 404 Advanced Threat Intelligence Team I. Introduction SilverFox has become one of the most active cyber threats in recent years, targeting managerial and finance staff in organization...",
"pubDate": "Fri, 13 Mar 2026 08:30:00 +0800",
"guid": "https://paper.seebug.org/3472",
"category": "威胁情报"
},
{
"title": "基于图像的提示注入:通过视觉嵌入的对抗性指令劫持多模态大语言模型",
"link": "https://paper.seebug.org/3471",
"description": "作者Neha Nagaraja, Lan Zhang, Zhilong Wang 译者知道创宇404实验室翻译组 原文链接https://arxiv.org/html/2603.03637v1/https://arxiv.org/html/2603.03637v1 摘要多模态大语言模型MLLMs融合视觉与文本能力赋能各类应用但这种融合也引入了新的安全漏洞。本文研究基于图像的提示注入...",
"pubDate": "Wed, 11 Mar 2026 01:29:00 +0800",
"guid": "https://paper.seebug.org/3471",
"category": "AI安全"
},
{
"title": "虚假 OpenClaw 安装程序如何传播 GhostSocks 恶意软件",
"link": "https://paper.seebug.org/3470",
"description": "作者Jai Minton, Ryan Dowd 原文链接https://www.huntress.com/blog/openclaw-github-ghostsocks-infostealer/https://www.huntress.com/blog/openclaw-github-ghostsocks-infostealer 摘要 信息窃取型恶意软件是针对面向公众系统发起严重攻击的初始访问...",
"pubDate": "Tue, 10 Mar 2026 01:45:00 +0800",
"guid": "https://paper.seebug.org/3470",
"category": "情报分析"
},
{
"title": "当 AI Agent 拥有系统权限OpenClaw 安全风险全景分析",
"link": "https://paper.seebug.org/3469",
"description": "作者:知道创宇积极防御实验室 完整的 OpenClaw 安全实践 Skill、自动化审计工具已在 GitHub 开源。项目地址https://github.com/knownsec/openclaw-security/https://github.com/knownsec/openclaw-security 一、背景说明 随着 AI Agent 技术的快速演进,具备自动化决策与自主执行能力的开源...",
"pubDate": "Fri, 06 Mar 2026 07:34:00 +0800",
"guid": "https://paper.seebug.org/3469",
"category": "AI安全"
}
]