AI Security AI安全 4d ago Updated 4d ago 更新于 4天前 41

Cavern C2 Uses DNS and Google Apps Script to Blend Into Legitimate Traffic Cavern C2 利用 DNS 和 Google Apps Script 融入合法流量

Cavern C2 framework, used by Iranian nation-state hackers (Cavern Manticore/MOIS), has evolved with a new module (GoogleService.dll) that uses DNS A-record queries to dynamically choose between direct HTTPS and Google Apps Script relays for C2 communication HOLLOWGRAPH module abuses Microsoft 365 calendars as covert two-way dead-drop C2 channels via Microsoft Graph API, with calendar events dated to 2050 to avoid detection The framework employs a modular plugin-based architecture with an inter-c Cavern C2框架新增GoogleService.dll模块,通过DNS A记录查询动态选择直接HTTPS或Google Apps Script中继进行通信,实现流量伪装 HOLLOWGRAPH模块将Microsoft 365日历转化为双向死投C2通道,利用Graph API以2050年日历事件形式外泄数据并接收指令 攻击者通过DNS隧道刷新Microsoft Entra ID凭证,结合.NET NativeAOT编译技术提升模块隐蔽性 APT42组织重新活跃,使用TAMECAT框架针对核能行业进行鱼叉式攻击,并系统性应用生成式AI加速攻击链各环节 Cavern框架向模块化插件架构演进,通

58
Hot 热度
65
Quality 质量
55
Impact 影响力

Analysis 深度分析

TL;DR

  • Cavern C2 framework, used by Iranian nation-state hackers (Cavern Manticore/MOIS), has evolved with a new module (GoogleService.dll) that uses DNS A-record queries to dynamically choose between direct HTTPS and Google Apps Script relays for C2 communication
  • HOLLOWGRAPH module abuses Microsoft 365 calendars as covert two-way dead-drop C2 channels via Microsoft Graph API, with calendar events dated to 2050 to avoid detection
  • The framework employs a modular plugin-based architecture with an inter-component broker (rnp.dll) supporting runtime upgrades and dynamic component loading
  • APT42 resurfaces with TAMECAT, a modular surveillance framework using AI-accelerated operations for spear-phishing targeting nuclear energy sector via LNK files disguised as PDFs
  • Both threats demonstrate sophisticated abuse of legitimate cloud services (Google Apps Script, Microsoft Graph API) to blend malicious traffic with normal network activity

Why It Matters

This highlights the growing trend of nation-state actors leveraging legitimate SaaS platforms as C2 infrastructure, making traditional perimeter-based detection increasingly ineffective. The integration of generative AI into offensive operations by groups like APT42 represents a significant escalation in operational tempo and capability, requiring defenders to adopt cloud-native detection strategies and enhanced monitoring of legitimate service APIs.

Technical Details

  • GoogleService.dll: Reads "conf.json" configuration, performs DNS A-record queries to select between direct HTTPS or Google Apps Script relay for each C2 transaction; same DNS infrastructure can validate and rotate relay deployment IDs
  • HOLLOWGRAPH: .NET NativeAOT-compiled DLL (first detected June 7, 2026) that uses Microsoft Graph API to treat compromised calendars as dead-drops; commands planted as calendar events dated May 13, 2050 with encrypted payloads; DNS tunneling refreshes Microsoft Entra ID credentials
  • rnp.dll: Inter-component broker functioning as local bridge for DLL discovery, message routing, and runtime upgrades within the Cavern framework
  • Cavern modules: File operations, SQL enumeration, Active Directory reconnaissance, LDAP brute-force, network reconnaissance, SOCKS5 proxy, and WebSocket tunneling
  • TAMECAT: Modular surveillance framework supporting enumeration, arbitrary command execution, browser credential/cookie collection, Outlook .ost mailbox harvesting, screenshot capture, with fallback C2 and exfiltration mechanisms

Industry Insight

  • Organizations should implement enhanced monitoring and anomaly detection for API calls to Microsoft Graph and Google Apps Script, particularly focusing on unusual calendar event patterns and credential refresh behaviors
  • The AI-accelerated operational model demonstrated by APT42 suggests defenders must invest in automated threat intelligence analysis and behavioral detection to keep pace with rapidly evolving attack tooling
  • Cloud service abuse as C2 infrastructure requires a shift from perimeter-based security to identity-centric and API-traffic monitoring strategies, including baseline behavioral analytics for legitimate SaaS platforms

TL;DR

  • Cavern C2框架新增GoogleService.dll模块,通过DNS A记录查询动态选择直接HTTPS或Google Apps Script中继进行通信,实现流量伪装
  • HOLLOWGRAPH模块将Microsoft 365日历转化为双向死投C2通道,利用Graph API以2050年日历事件形式外泄数据并接收指令
  • 攻击者通过DNS隧道刷新Microsoft Entra ID凭证,结合.NET NativeAOT编译技术提升模块隐蔽性
  • APT42组织重新活跃,使用TAMECAT框架针对核能行业进行鱼叉式攻击,并系统性应用生成式AI加速攻击链各环节
  • Cavern框架向模块化插件架构演进,通过滥用合法云服务(Google/Microsoft)规避传统边界检测

为什么值得看

本文揭示了国家级APT组织将合法云服务深度嵌入C2架构的最新战术演进,为防御者提供了识别隐蔽通信的新指标。同时展示了生成式AI如何被系统性整合到攻击链中,对威胁情报分析和安全架构设计具有直接参考价值。

技术解析

  • DNS驱动的双通道C2机制:GoogleService.dll模块读取conf.json配置,通过DNS A记录查询结果动态决定每次通信使用直接HTTPS连接还是Google Apps Script中继,同一DNS基础设施可同时验证和轮换中继部署ID
  • 日历事件死投通信:HOLLOWGRAPH利用Microsoft Graph API将受害邮箱日历转化为双向C2通道,攻击者以2050年5月13日的未来日期创建日历事件植入指令,被控主机通过创建带加密附件的日历事件外泄数据
  • 凭证维持技术:采用DNS隧道技术持续刷新Microsoft Entra ID OAuth刷新令牌,并将更新后的凭证写入磁盘文本文件,确保长期访问能力
  • 模块化架构演进:Cavern框架于2026年4月底转向插件式架构,rnp.dll作为组件间通信桥梁实现DLL动态加载、消息路由和运行时升级,支持SOCKS5代理、WebSocket隧道等扩展模块
  • AI辅助攻击链:APT42在鱼叉式攻击中系统性应用生成式AI,用于开发专用工具、研究漏洞利用技术、多语言翻译、目标邮箱识别和实体情报收集

行业启示

  • 合法云服务滥用成为C2新范式:攻击者将Google Workspace、Microsoft 365等可信服务深度嵌入通信架构,传统基于域名信誉和流量异常检测的防御手段失效,需建立云服务API行为基线监控
  • AI赋能攻击链的防御应对:当生成式AI被整合到目标研究、工具开发和社交工程环节时,威胁情报需关注AI生成内容的特征指标,同时加强针对高价值目标的人工验证机制
  • 模块化框架的持续演进风险:Cavern等框架的快速迭代能力表明,单一漏洞修补不足以应对国家级APT,需要建立覆盖通信协议、API滥用模式和凭证生命周期的纵深防御体系

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

Security 安全 Research 科学研究