Aggregator
CVE-2024-27856 | Apple macOS up to 17.4 File denial of service (Nessus ID 216648)
CVE-2024-40771 | Apple macOS memory corruption
CVE-2024-40771 | Apple tvOS memory corruption
CVE-2024-40771 | Apple visionOS memory corruption
CVE-2024-40771 | Apple iOS/iPadOS memory corruption
CVE-2024-40771 | Apple watchOS memory corruption
CVE-2024-40854 | Apple macOS initialization (WID-SEC-2024-3291)
Simplify Your Approach to Securing OT Networks
Securing OT networks isn't about eliminating risk. It's about managing it strategically. Learn how a three-pillar framework of risk assessment, tolerance and acceptance, paired with a phased approach to microsegmentation, can turn an overwhelming challenge into a manageable journey.
Regulation Didn’t Change, Your Identity Landscape Did
Mass. Hospital Diverting Ambulances as It Deals With Attack
A Massachusetts healthcare system is diverting ambulance patients and is operating under downtime procedures as it deals with a cyberattack. The organization has also canceled certain cancer treatments, taken its patient portal offline and is unable to fill prescriptions at its retail pharmacies.
New eSentire CEO Pursues AI-Driven Managed Security Shift
CEO James Foster says managed detection and response is evolving into an AI-powered agentic model as enterprises face faster AI-driven threats. He stresses balancing automation with human expertise while positioning eSentire as a vendor-neutral platform integrating best-of-breed security tools.
US Critical Infrastructure Facing Iranian-Linked OT Threats
Federal agencies are warning that Iranian-linked actors have begun actively exploiting internet-facing PLCs and misconfigured OT systems across U.S. critical infrastructure, enabling network access, lateral movement and potential disruption amid rising geopolitical tensions.
Anthropic Calls Its New Model Too Dangerous to Release
Anthropic asserted Tuesday that it's created a new era for cybersecurity after developing an artificial intelligence model too dangerous to release to public. The company's unreleased Claude Mythos Preview model has already found thousands of high-severity vulnerabilities.
Квантовый взлом перестал быть теорией. Ученые нашли способ обрушить безопасность интернета в ближайшие годы
JVN: Xerox FreeFlow Coreにおける複数の脆弱性(XRX26-005)
JVN: Xerox FreeFlow Coreにおける複数の脆弱性
Claude Code源码泄露遭利用,攻击者借GitHub散播窃密木马
Claude Code源码泄露遭利用,攻击者借GitHub散播窃密木马
威胁组织正利用近期Claude Code源代码泄露事件,通过伪造GitHub仓库向用户分发Vidar窃密木马。
Claude Code是人工智能公司Anthropic推出的一款终端版AI代理工具,可直接在终端中执行编程任务,作为自主代理实现系统直接交互、大语言模型API调用管理、MCP集成以及持久化记忆等功能。
据悉,Anthropic在发布npm包时因疏忽,意外嵌入了一个大小为59.8MB的JavaScript源码映射文件,导致这款新工具的完整客户端源代码被公开泄露。
此次泄露包含1906个文件、共计51.3万行未做混淆处理的TypeScript代码,暴露了该AI代理的调度逻辑、权限与执行机制、隐藏功能、编译细节以及多项安全相关核心实现。
泄露代码很快被大量用户下载并上传至GitHub,出现数千次复刻分叉。
Zscaler在一份报告中指出,此次源码泄露为威胁组织创造了可乘之机,使其能够向搜索Claude Code泄露内容的用户投放Vidar窃密软件。
研究人员发现,一个由用户“idbzoomh”发布的恶意GitHub仓库上传了伪造的泄露版本,并宣称该版本“解锁企业级功能”且无使用限制。
GitHub代码库传播恶意软件
为吸引更多流量,该仓库针对搜索引擎做了优化,在谷歌搜索“leaked Claude Code”等关键词时会出现在前列位置。
恶意GitHub代码库的搜索结果
据研究人员分析,用户下载到一个7-Zip压缩包,其中包含一个名为ClaudeCode_x64.exe的Rust语言可执行程序。该程序运行后,会释放投放通用窃密木马Vidar,同时安装GhostSocks网络流量代理工具。
Zscaler监测发现,该恶意压缩包会频繁更新,未来不排除被加入其他恶意载荷的可能。
研究人员还发现了第二个代码完全相同的GitHub仓库,但该仓库中的“下载ZIP”按钮在分析期间处于失效状态。Zscaler判断,该仓库由同一威胁组织运营,很可能用于测试不同的投放传播策略。
第二个与同一威胁者关联的GitHub代码库
尽管GitHub平台具备安全防护机制,仍频繁被攻击者用于以各种伪装方式分发恶意载荷。
在2025年末的多轮攻击活动中,威胁组织就曾针对经验不足的研究人员或黑客,在仓库中谎称存放最新披露漏洞的概念验证(PoC)利用代码。
从以往案例来看,攻击者往往会迅速借热点公开事件发动机会主义攻击,借机入侵目标设备。