AI Security AI安全 8h ago Updated 1h ago 更新于 1小时前 43

CISA Red Team Compromised Two Critical Infrastructure Orgs, One Detected Nothing CISA红队入侵两家关键基础设施组织,一家未检测到任何异常

CISA conducted simultaneous red team assessments against two critical infrastructure organizations using identical tradecraft, revealing dramatically different defensive outcomes despite similar vulnerabilities. Organization A (Government Services) was fully compromised with zero detections due to alert fatigue, siloed SOCs, and poor operational processes. Organization B (Water/Wastewater) detected and contained initial access within 2-20 minutes, demonstrating that people and processes matter m CISA对两个关键基础设施组织进行同步红队评估,使用相似战术却得出截然不同的防御结果 Organization A(政府服务设施部门)完全被攻破且零检测,存在默认凭据、AD CS配置错误、明文存储凭据、永不过期的云密钥等严重问题 Organization B(水务部门)SOC在2-20分钟内检测到初始钓鱼载荷并隔离工作站,成功阻断攻击传播 两个组织存在相同的基础设施安全缺陷,但人员流程差异导致结果天壤之别 CISA强调"检测工具的有效性取决于支持它们的人员、流程和程序"

68
Hot 热度
62
Quality 质量
55
Impact 影响力

Analysis 深度分析

TL;DR

  • CISA conducted simultaneous red team assessments against two critical infrastructure organizations using identical tradecraft, revealing dramatically different defensive outcomes despite similar vulnerabilities.
  • Organization A (Government Services) was fully compromised with zero detections due to alert fatigue, siloed SOCs, and poor operational processes.
  • Organization B (Water/Wastewater) detected and contained initial access within 2-20 minutes, demonstrating that people and processes matter more than tools.
  • Both organizations shared critical misconfigurations: default Machine Account Quota, misconfigured AD CS templates (ESC1), cleartext credentials, static cloud keys, and over-permissioned Entra ID applications.
  • CISA's advisory AA26-237A "A Tale of Two SOCs" emphasizes that detection effectiveness depends on the operational framework supporting security tools.

Why It Matters

This case study provides a stark, real-world demonstration that security tooling alone cannot protect critical infrastructure—organizational processes, analyst training, and incident response procedures are equally critical. For AI and security practitioners, it underscores the importance of tuning detection systems to reduce false positives and ensuring SOC analysts have clear escalation paths and authority to act. The findings are directly applicable to any organization managing Active Directory, cloud identities, or operational technology environments.

Technical Details

  • Initial Access Vector: Both organizations were targeted with phishing emails sent from internal addresses via a web application using default credentials for built-in accounts.
  • Privilege Escalation: The red team exploited a default Machine Account Quota combined with a misconfigured AD CS certificate template (ESC1), enabling certificate requests for any user—a technique related to the Certighost domain-takeover exploit.
  • Credential Access: Cleartext credentials were found in decrypted database configuration files and static AWS access keys configured to never expire, with no token revocation mechanism in place.
  • Cloud Compromise: In Organization A, the team stole a Primary Refresh Token and abused over-permissioned Entra ID applications to read security team emails and assess detection awareness.
  • Organization B Response: After initial phishing payloads were detected and workstations isolated within 2-20 minutes, CISA shifted to an assume-breach model, replicating access on a designated non-privileged host and confirming the same underlying vulnerabilities existed.

Industry Insight

  • Alert Fatigue Is a Critical Vulnerability: Organization A's thousands of false-positive alerts from normal business operations drowned out genuine threats. Security teams must prioritize alert triage, implement noise-reduction strategies, and ensure high-severity ratings are reserved for truly critical events.
  • Siloed Security Operations Undermine Defense: Multiple SOCs and endpoint tools with no shared visibility allowed the red team to operate undetected. Organizations should invest in unified security platforms and cross-SOC coordination to eliminate blind spots.
  • People and Processes Outweigh Tooling: CISA's core finding—that the same tools produced different outcomes based on operational maturity—should drive investment in analyst training, clear escalation procedures, and defined authority for SOC teams to respond without bureaucratic delay.

TL;DR

  • CISA对两个关键基础设施组织进行同步红队评估,使用相似战术却得出截然不同的防御结果
  • Organization A(政府服务设施部门)完全被攻破且零检测,存在默认凭据、AD CS配置错误、明文存储凭据、永不过期的云密钥等严重问题
  • Organization B(水务部门)SOC在2-20分钟内检测到初始钓鱼载荷并隔离工作站,成功阻断攻击传播
  • 两个组织存在相同的基础设施安全缺陷,但人员流程差异导致结果天壤之别
  • CISA强调"检测工具的有效性取决于支持它们的人员、流程和程序"

为什么值得看

本文通过真实红队对比案例揭示了关键基础设施领域安全运营的核心痛点:技术工具并非决定性因素,人员响应能力和流程完善程度才是防御成败的关键。对AI安全从业者而言,这提供了关于威胁检测、SOC运营和身份安全建设的实战参考。

技术解析

  • 初始访问与权限提升:红队利用Web应用内置账户的默认凭据发送钓鱼邮件,通过默认Machine Account Quota和配置错误的AD CS证书模板(ESC1)实现权限提升,与近期披露的Certighost域接管漏洞属同类技术
  • 凭据窃取与云渗透:攻击者获取明文存储的数据库配置和永不过期的静态AWS访问密钥,窃取Primary Refresh Token并利用过度授权的Entra ID应用读取安全团队邮件以侦察防御状态
  • Organization A的防御失效:数千条来自正常业务的高严重性误报警报掩盖了红队活动,多个SOC和端点工具缺乏共享可见性,分析师缺乏升级流程和行动权限,一条真实的SCCM服务器警报因无法识别系统所有者而被当作误报忽略
  • Organization B的有效响应:SOC检测到初始钓鱼载荷执行并隔离工作站,切断C2通信;红队转为假设入侵模型,在指定非特权主机上执行payload,仍发现SCCM配置文件中域服务账户的明文凭据(拥有DC权限),成功执行DCSync获取krbtgt密钥
  • OT环境边界:Organization B的OT DMZ堡垒主机阻止了出站互联网访问,未建立C2通道,红队未能进入OT系统

行业启示

  • SOC运营质量比工具堆砌更重要:Organization A拥有多重安全工具和SOC却零检测,说明缺乏统一可见性、清晰升级流程和分析师授权会导致工具投资失效
  • 身份安全是关键防线:两个案例均暴露AD配置错误、明文凭据存储、过度授权应用等身份相关问题,强化身份治理和凭据保护是防御纵深的基础
  • 假设入侵模式的价值:Organization B在初始访问被阻断后,CISA转为假设入侵模型验证纵深防御能力,这种模式能更真实地评估安全体系的残余风险

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