安全日报 - 2026-09-05 - Cyber Security Daily Radar

安全公告2026-09-04 发布INFO

In the Linux kernel, the following vulnerability has been resolved: erofs: cap LZMA stream pool...

CVE-2026-80892

In the Linux kernel, the following vulnerability has been resolved: erofs: cap LZMA stream pool size fs/erofs/decompre

历史摘要 · 未重新核验 · GitHub Advisory

#advisory

安全公告2026-09-04 发布INFO

In the Linux kernel, the following vulnerability has been resolved: netfs: release readahead...

CVE-2026-80897

In the Linux kernel, the following vulnerability has been resolved: netfs: release readahead folios on iterator prepara

历史摘要 · 未重新核验 · GitHub Advisory

#advisory

安全公告2026-09-04 发布INFO

CodeWhale: rlm_eval auto-approves arbitrary Python execution, bypassing the user's approval policy (RCE)

CVE-2026-75858

### Maintainer resolution The CodeWhale maintainers validated this report. The affected package ranges are recorded in

历史摘要 · 未重新核验 · GitHub Advisory

#advisory #rust #npm #cwe-94 #cwe-862 #severity:high

安全公告2026-09-04 发布INFO

CodeWhale: js_execution leaks parent environment to model context via missing env scrub

CVE-2026-75915

### Maintainer resolution The CodeWhale maintainers validated this report. The affected package ranges are recorded in

历史摘要 · 未重新核验 · GitHub Advisory

#advisory #rust #npm #cwe-200 #cwe-526 #severity:high

安全公告2026-09-03 发布INFO

SiYuan: Absolute filesystem path and OS username disclosure via resolveAssetPath

CVE-2026-72802

**CVE:** This vulnerability corresponds to [CVE-2026-72802](https://nvd.nist.gov/vuln/detail/CVE-2026-72802). ### Summa

历史摘要 · 未重新核验 · GitHub Advisory

#advisory #go #cwe-639 #severity:medium

学术论文2026-09-02 发布INFO

SEAL: Reinforcing Global Safety in Mixture-of-Experts through Shared Expert ALignment

本文针对混合专家（MoE）大语言模型在安全对齐方面的结构性缺陷提出了一种新的防御方法。MoE 架构通过稀疏路由仅激活少量专家模块，从而在近乎恒定的计算开销下实现海量参数扩展；近期出现的混合 MoE 进一步引入共享专家（shared expe

历史摘要 · 未重新核验 · arXiv cs.CR

#cs.LG #cs.AI #agent-security #llm #mixture-of-experts #safety-alignment

学术论文2026-09-01 发布INFO

Athena: Vulnerability-Affected Library Identification via Knowledge Graph Completion

该论文提出 Athena，一个基于知识图谱的漏洞影响库识别系统。研究背景是：广泛使用的库中单个漏洞可能影响数百万下游应用，但超过半数的漏洞数据库条目存在受影响库信息缺失或错误的问题。现有自动化方法将识别问题当作孤立的文本检索任务，忽略了漏洞

历史摘要 · 未重新核验 · arXiv cs.CR

#cs.SE #cs.AI #agent-security #llm #knowledge-graph #vulnerability-database

学术论文2026-09-03 发布INFO

SENTINEL-RL: Offloading Topological Reasoning from LLM Agents in the Security Operations Center

该论文针对大语言模型（LLM）代理在安全运营中心（SOC）中作为自主分析师时面临的两个核心可扩展性缺陷展开研究：一是有限的上下文窗口无法容纳包含数千台主机的认证图全貌；二是自由文本生成无法保证推荐的安全处置动作与当前网络拓扑结构在逻辑上一致

历史摘要 · 未重新核验 · arXiv cs.CR

#cs.AI #agent-security #llm #soc #graph-neural-network

学术论文2026-09-03 发布INFO

Trust Me, I'm Your Developer: Self-Issued Authentication in Large Language Models

大型语言模型（LLM）的安全研究大多集中在角色扮演越狱等攻击，忽略了当用户要求模型通过模型自身设计的测试来验证身份声明时的行为。本研究针对这一缺口进行了分阶段开发者身份实验，涉及 ChatGPT、Claude、Qwen、Mistral 和

历史摘要 · 未重新核验 · arXiv cs.CR

#llm-security #authentication #identity #research

学术论文2026-09-02 发布INFO

Privacy-Preserving Topology-Guided Safety for LLM-Based Multi-Agent Systems via Federated Graph Learning

本文针对基于大语言模型的多智能体系统（LLM-based MAS）提出一种隐私保护的安全防护框架。现有基于拓扑的防护方法通常通过图神经网络（GNN）在智能体间通信图上训练，以定位风险智能体并干预拓扑，但这类方法假设一个中心化运营商能聚合所有

历史摘要 · 未重新核验 · arXiv cs.CR

#cs.AI #cs.LG #cs.MA #agent-security #llm #federated-learning

威胁情报2026-09-05 发布INFO

Cloud & Application Security

历史摘要 · 未重新核验 · CrowdStrike

#edr #apt #crowdstrike

威胁情报2026-09-05 发布INFO

From Scanner to Stealer: Inside the trivy-action Supply Chain Compromise

历史摘要 · 未重新核验 · CrowdStrike

#edr #apt #crowdstrike

威胁情报2026-09-05 发布INFO

Next-Gen Identity Security

历史摘要 · 未重新核验 · CrowdStrike

#edr #apt #crowdstrike

威胁情报2026-09-05 发布INFO

CrowdStrike Expands Identity Leadership with OpenID and IDPro

历史摘要 · 未重新核验 · CrowdStrike

#edr #apt #crowdstrike

威胁情报2026-09-05 发布INFO

cloud architecture assessment

历史摘要 · 未重新核验 · Mandiant

#report #mandiant #apt

威胁情报2026-09-05 发布INFO

ransomware and multifaceted extortion defense

历史摘要 · 未重新核验 · Mandiant

#report #mandiant #apt

威胁情报2026-09-05 发布INFO

Google Cloud Next 2026Catch-up sessions on the keynotes and select sessions are now available on demand.Explore content on-demand

Google Cloud Next 2026Catch-up sessions on the keynotes and select sessions are now available on demand.Explore content

历史摘要 · 未重新核验 · Mandiant

#report #mandiant #apt

威胁情报2026-09-05 发布INFO

The Identity Problem Hiding in AI Agent Deployments

历史摘要 · 未重新核验 · CrowdStrike

#edr #apt #crowdstrike

威胁情报2026-09-05 发布INFO

94% of Organizations Report Cloud Breaches: CrowdStrike State of CDR Survey

历史摘要 · 未重新核验 · CrowdStrike

#edr #apt #crowdstrike

威胁情报2026-09-05 发布INFO

Falcon Cloud Security June 2026 Release: Updates for Azure and Google Cloud

历史摘要 · 未重新核验 · CrowdStrike

#edr #apt #crowdstrike

2026-09-05 归档