UI misrepresentation in CustomTabs in Google Chrome on on Android prior to 153.0.8010.36 allowed...
CVE-2026-87597
UI misrepresentation in CustomTabs in Google Chrome on on Android prior to 153.0.8010.36 allowed a remote attacker to sp
2026-09-09 归档
今日共收录 1483 条安全情报，包含 593 个 CVE，506 篇安全通告，以及 97 篇研究论文。
← 返回归档列表Improper input validation in Safebrowsing in Google Chrome on on Android prior to 153.0.8010.36...
CVE-2026-87600
Improper input validation in Safebrowsing in Google Chrome on on Android prior to 153.0.8010.36 allowed a remote attacke
Server-side request forgery in Mobile in Google Chrome prior to 153.0.8010.36 allowed a remote...
CVE-2026-87595
Server-side request forgery in Mobile in Google Chrome prior to 153.0.8010.36 allowed a remote attacker leveraging socia
Windows ML CLI: CORS misconfig enables localhost RCE
CVE-2026-84452
Case Description: MSRC Notes: Attachments: 1 file(s) attached (1 mp4) Summary: The vulnerability lies in the 'serve/cli
Hono: Query parser reads parameters after the URL fragment, causing cache-key and proxy interpretation differentials
CVE-2026-84363
### Summary Hono's query parsing does not stop at the URL fragment: a `?` appearing after a `#` is treated as the start
SVGO: removeScripts allows executable links through namespace and control-character bypasses
CVE-2026-84370
## Summary SVGO's opt-in `removeScripts` plugin failed to remove some executable links. Namespace-prefixed SVG anchors
SVGO: removeScripts incompletely sanitizes executable HTML in SVG foreignObject elements
CVE-2026-84369
## Summary SVGO's opt-in `removeScripts` plugin did not inspect executable HTML content inside SVG `` elements. Applica
xmldom: Processing Instruction Target Injection Bypasses requireWellFormed
CVE-2026-83616
## Summary `Document.createProcessingInstruction()` in `@xmldom/xmldom` performs no validation on the `target` paramete
HoneyRoute: Honeypot-Model Routing for Adversarial LLM Serving
本文提出了一种名为 HoneyRoute 的推理服务层，用于在面向对抗性的大语言模型（LLM）服务中检测恶意请求并将其路由到专用蜜罐模型，从而保护生产模型，同时持续收集攻击者交互并转化为情报。现有防御通常将陷阱嵌入模型内存或在协议层重建欺骗
AgentLeak: Cloning Stronger LLM Agent Capabilities onto Weaker Agents Beyond Skill Stealing
本文提出一种针对大语言模型（LLM）智能体的新型黑盒能力克隆攻击方法 AgentLeak。研究背景是：LLM 智能体通过基础模型与显式技能（skills）及执行中获得的隐式程序性知识相结合，能够完成长期任务，这种任务解决能力已成为宝贵的商业
Structurally Close, Temporally Distant: Measuring Security Exposure in Long-Horizon LLM Agents
长期运行的大语言模型（LLM）智能体在与不受信任的内容、持久化内存、外部状态和敏感工具交互时，面临严重的安全风险。现有的安全分析往往以恶意输入与下游动作之间的执行步数来衡量攻击的距离，但本文指出，对于有状态的智能体，这种时间上的距离可能高估
Measuring the Security of the Evolving Software Supply Chain: a Research Agenda
软件供应链安全已成为现代软件工程中的核心关注点，因为现代软件对第三方依赖有着广泛依赖，且攻击面持续扩大。然而，现有的基于量化测量的分析方法和漏洞管理手段往往碎片化，并局限于特定生态系统（如npm、PyPI等），难以在跨环境之间提供可比较的风
Evidence-Grounded Retrieval for Investigation Hunt Lead Generation from CTI Reports
该论文针对网络威胁猎杀中依赖人工将威胁情报（CTI）报告转化为可执行调查线索（hunt leads）的低效问题，提出了一种名为 AHLERT 的自动化系统。现有方法大多停留在实体提取层面，忽略防御者自身的运行环境，且孤立分析每份报告。AHL
Do AI Coding Assistants Check Before They Install? A Pre-Registered Demand-Side Audit of Trust Signals in the Research Software Supply Chain
本研究针对AI编程助手在自动选择、安装和配置软件时是否利用供应链安全信号的问题，进行了预注册的受控实证审计。背景是攻击者已通过伪造包名、入侵维护者账号、篡改仓库文本等方式利用助手的能力实施攻击；供应链社区为此发布了机器可验证的信任信号，包括
Staying on the Attack Path: Structured State for Long-Horizon Automated Penetration Testing
本文针对基于大型语言模型（LLM）的智能体在长周期自动化渗透测试任务中面临的上下文遗忘和意图漂移问题，提出了一种名为Intentest的意图图引导的自动化渗透测试智能体。在长时间交互中，早期发现的关键事实和因果推理链容易丢失，导致智能体陷入
Scratchy: Visual-Scratchpad Multimodal Reasoning for Cryptographic Proof Generation in EasyCrypt
该论文提出了一种名为 Scratchy 的视觉暂存器多模态推理方法，旨在解决大语言模型（LLM）在生成密码学形式化证明时面临的结构性依赖难题。在密码学领域，计算安全性论证要求证明必须协调概率、对抗游戏、不变量、假设和界限等要素，而 Easy
Heap-based buffer overflow in Windows NTFS allows an unauthorized attacker to execute code over a network.
CVE-2026-69463
CVE-2026-69463 是位于 Windows NTFS 组件中的一个堆缓冲区溢出漏洞。根据 NVD 提供的元数据，该漏洞可使未经授权的攻击者通过网络远程触发，并有可能执行任意代码。CVSS 3.1 基础评分为 9.8，属于严重程度（
Cloud & Application Security
Cloud & Application Security
From Scanner to Stealer: Inside the trivy-action Supply Chain Compromise
From Scanner to Stealer: Inside the trivy-action Supply Chain Compromise
Next-Gen Identity Security
Next-Gen Identity Security