AI Security AI安全 7h ago Updated 4h ago 更新于 4小时前 49

Exploitation of ServiceNow Vulnerability Seen Days After Disclosure 披露数天后发现ServiceNow漏洞被利用

A critical remote code execution vulnerability (CVE-2026-6875) in the ServiceNow AI platform is being actively exploited in the wild. The flaw functions as a sandbox escape, allowing unauthenticated attackers to execute arbitrary code on affected systems. ServiceNow deployed patches for hosted instances on July 14, but self-hosted customers must manually apply the update. Threat intelligence firm Defused confirmed in-the-wild exploitation, noting the payload matched a proof-of-concept released b ServiceNow AI平台存在关键远程代码执行漏洞CVE-2026-6875,属于沙箱逃逸问题,允许未认证攻击者执行任意代码。 威胁情报公司Defused于7月18日报告在野外观察到该漏洞的利用活动,且载荷与Searchlight Cyber发布的概念验证一致。 ServiceNow已发布补丁,托管实例已自动更新,但自托管客户需手动安装,厂商目前尚未确认其托管环境受到直接影响。 尽管存在利用报告,但尚无证据表明这是由恶意攻击者发起的大规模攻击,不排除是安全研究人员进行的测试行为。 ServiceNow漏洞被真实威胁行为者利用的情况较为罕见,CISA KEV目录中仅收录了两个2024年修复的

75
Hot 热度
65
Quality 质量
70
Impact 影响力

Analysis 深度分析

TL;DR

  • A critical remote code execution vulnerability (CVE-2026-6875) in the ServiceNow AI platform is being actively exploited in the wild.
  • The flaw functions as a sandbox escape, allowing unauthenticated attackers to execute arbitrary code on affected systems.
  • ServiceNow deployed patches for hosted instances on July 14, but self-hosted customers must manually apply the update.
  • Threat intelligence firm Defused confirmed in-the-wild exploitation, noting the payload matched a proof-of-concept released by Searchlight Cyber.

Why It Matters

This incident highlights the severe risks associated with sandbox escapes in enterprise AI platforms, where isolation failures can lead to full system compromise. It underscores the urgent need for organizations using self-hosted ServiceNow instances to prioritize immediate patching to prevent unauthorized code execution.

Technical Details

  • Vulnerability Type: Critical Remote Code Execution (RCE) via sandbox escape.
  • Impact: Allows unauthenticated attackers to execute arbitrary code on the target infrastructure.
  • Patch Status: Patches were released on July 14; applied automatically for hosted instances, requiring manual installation for self-hosted environments.
  • Exploit Analysis: Initial reports suggested varied exploitation methods, but subsequent analysis by Defused confirmed the in-the-wild payload was identical to the proof-of-concept disclosed by Searchlight Cyber.

Industry Insight

  • Organizations relying on self-hosted enterprise software must implement rigorous and rapid patch management protocols to mitigate exposure to newly disclosed critical vulnerabilities.
  • The discrepancy between vendor advisories stating "no known exploitation" and independent threat intelligence confirming active attacks demonstrates the importance of cross-referencing multiple security sources during incident response.
  • Sandbox integrity is a critical security boundary in AI platforms; vendors and users alike must treat sandbox escape vulnerabilities with the highest severity due to their potential for complete system takeover.

TL;DR

  • ServiceNow AI平台存在关键远程代码执行漏洞CVE-2026-6875,属于沙箱逃逸问题,允许未认证攻击者执行任意代码。
  • 威胁情报公司Defused于7月18日报告在野外观察到该漏洞的利用活动,且载荷与Searchlight Cyber发布的概念验证一致。
  • ServiceNow已发布补丁,托管实例已自动更新,但自托管客户需手动安装,厂商目前尚未确认其托管环境受到直接影响。
  • 尽管存在利用报告,但尚无证据表明这是由恶意攻击者发起的大规模攻击,不排除是安全研究人员进行的测试行为。
  • ServiceNow漏洞被真实威胁行为者利用的情况较为罕见,CISA KEV目录中仅收录了两个2024年修复的漏洞。

为什么值得看

对于使用ServiceNow平台的AI从业者和企业IT管理者而言,此事件凸显了AI集成平台在引入新功能时可能伴随的安全风险,特别是沙箱逃逸这类高危漏洞。它提醒行业在部署第三方AI服务时,必须严格区分托管与自托管环境的维护责任,并密切关注供应商的安全响应速度及实际利用情况。

技术解析

  • 漏洞性质:CVE-2026-6875被定义为关键级别的远程代码执行(RCE)漏洞,具体机制为沙箱逃逸。这意味着攻击者可以在受限环境中执行代码,从而突破隔离限制,在未授权情况下控制服务器。
  • 利用现状:网络安全公司Searchlight Cyber披露了技术细节和概念验证(PoC)。随后,Defused报告捕获了野外利用流量,初期认为方法不同,后经修正确认为相同的PoC载荷,证实了漏洞的可利用性。
  • 补丁部署差异:ServiceNow在7月14日发布补丁。对于云托管实例,服务商已主动部署更新;但对于自托管(Self-hosted)客户,必须自行应用补丁,这构成了潜在的安全滞后窗口。
  • 厂商响应状态:ServiceNow官方声明知晓相关利用活动,但调查至今未发现其托管实例受影响的证据。厂商建议所有客户检查并应用补丁,同时提供协助支持。

行业启示

  • 强化自托管环境的安全运维:对于依赖自托管AI平台的企业,必须建立快速响应机制,确保在漏洞披露后能立即应用补丁,避免因维护延迟导致被利用。
  • 警惕“研究性”利用背后的真实风险:虽然当前利用可能源于安全研究,但公开的PoC和详细技术细节会迅速降低攻击门槛,威胁行为者可能很快跟进,因此不能因“非恶意”标签而放松防御。
  • 重视AI平台的供应链安全:随着AI功能嵌入主流企业软件(如ServiceNow),传统的边界防御已不足够,需加强对平台内部组件(如沙箱机制)的安全性评估和持续监控。

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