AI Security AI安全 1d ago Updated 1d ago 更新于 1天前 46

Critical GitLab Flaw Exploited Shortly After Disclosure GitLab关键漏洞在披露后不久即遭利用

CVE-2026-19478 is a critical (CVSS 9.4) unauthenticated code injection vulnerability in GitLab that allows remote attackers to modify or delete public projects and user data via a GraphQL directive Exploitation began within approximately two days of public disclosure on August 17, 2026, with WatchTowr confirming in-the-wild attacks detected via honeypot networks GitLab released emergency patches across CE and EE versions (19.2.4, 19.1.6, 19.0.8, and 18.11.11), and recommended restricting unauthe GitLab披露关键漏洞CVE-2026-19478(CVSS 9.4),允许未认证攻击者通过GraphQL远程修改或删除公共项目及用户数据 漏洞披露后仅两天,WatchTowr蜜罐网络即捕获真实世界利用尝试,警告AI赋能攻击者可能紧随其后 补丁已发布至GitLab CE/EE版本19.2.4、19.1.6、19.0.8和18.11.11 该漏洞可伪造合并记录,可能引发新一轮供应链攻击浪潮 缓解措施包括限制/api/graphql端点未认证访问或移除公共仓库访问权限

72
Hot 热度
62
Quality 质量
58
Impact 影响力

Analysis 深度分析

TL;DR

  • CVE-2026-19478 is a critical (CVSS 9.4) unauthenticated code injection vulnerability in GitLab that allows remote attackers to modify or delete public projects and user data via a GraphQL directive
  • Exploitation began within approximately two days of public disclosure on August 17, 2026, with WatchTowr confirming in-the-wild attacks detected via honeypot networks
  • GitLab released emergency patches across CE and EE versions (19.2.4, 19.1.6, 19.0.8, and 18.11.11), and recommended restricting unauthenticated access to /api/graphql as an interim mitigation
  • The vulnerability enables forging merge records without credentials or user interaction, raising severe supply chain attack concerns as malicious changes can appear legitimately reviewed and signed off
  • AI-enabled attackers are expected to accelerate exploitation timelines, narrowing the already critical patch window for organizations

Why It Matters

This vulnerability represents a paradigm shift in supply chain security: the ability to forge merge records means attackers can make malicious code appear fully reviewed and approved, bypassing one of the most fundamental defenses in software development workflows. For AI practitioners and security teams, it underscores the accelerating pace of vulnerability exploitation in the AI-augmented threat landscape and the urgent need for rapid patching cadences and zero-trust access controls on critical infrastructure endpoints.

Technical Details

  • Vulnerability: CVE-2026-19478, a code injection flaw in GitLab's GraphQL API allowing unauthenticated remote code execution via the @gl_introduced directive, with a CVSS score of 9.4
  • Attack vector: A single HTTP request without any credentials, user interaction, or obscure configuration can delete repositories, forge merge records, or ban maintainers on publicly accessible GitLab instances
  • Patched versions: GitLab Community Edition (CE) and Enterprise Edition (EE) versions 19.2.4, 19.1.6, 19.0.8, and 18.11.11; original patch released August 17, 2026
  • Detection guidance: Organizations should hunt web logs for requests containing @gl_introduced to identify probing or exploitation attempts; WatchTowr reproduced the vulnerability within minutes using only advisory details and the patch diff
  • Mitigation: Restrict unauthenticated access to the /api/graphql endpoint or remove public repository access entirely until patches are applied

Industry Insight

  • The two-day exploitation window reinforces that AI-augmented threat actors are compressing the time between disclosure and active exploitation, making automated patch management and continuous vulnerability scanning non-negotiable for any organization running GitLab
  • The ability to forge merge records transforms this from a data destruction risk into a supply chain poisoning vector—organizations should audit recent merge activity on exposed instances and implement immutable audit logging with cryptographic verification for critical pipelines
  • The broader implication is that GraphQL endpoints on public-facing infrastructure represent an expanding attack surface; security teams should conduct immediate inventory and access audits of all GraphQL APIs, treating unauthenticated write operations as critical-risk findings

TL;DR

  • GitLab披露关键漏洞CVE-2026-19478(CVSS 9.4),允许未认证攻击者通过GraphQL远程修改或删除公共项目及用户数据
  • 漏洞披露后仅两天,WatchTowr蜜罐网络即捕获真实世界利用尝试,警告AI赋能攻击者可能紧随其后
  • 补丁已发布至GitLab CE/EE版本19.2.4、19.1.6、19.0.8和18.11.11
  • 该漏洞可伪造合并记录,可能引发新一轮供应链攻击浪潮
  • 缓解措施包括限制/api/graphql端点未认证访问或移除公共仓库访问权限

为什么值得看

本文揭示了AI时代漏洞利用速度的显著加速——关键漏洞披露后仅两天即被实际利用,安全响应窗口正在急剧收窄。同时,该漏洞的供应链攻击潜力意味着攻击者可以伪造代码审查记录,传统安全控制措施面临严峻挑战。

技术解析

  • 漏洞编号CVE-2026-19478,CVSS评分9.4,属于代码注入缺陷,通过GraphQL指令实现未认证远程代码执行
  • 补丁版本包括CE/EE的19.2.4、19.1.6、19.0.8和18.11.11,攻击者可通过单个HTTP请求删除仓库、伪造合并记录或封禁维护者
  • WatchTowr在披露后几分钟内即复现漏洞,其蜜罐网络已捕获真实利用尝试,检测指标包括请求中的"@gl_introduced"字符串
  • 缓解措施包括限制/api/graphql端点的未认证访问或完全移除公共仓库访问权限

行业启示

  • AI赋能的攻击者正在大幅缩短漏洞利用窗口,从披露到实际利用的时间已压缩至数天甚至数小时,安全团队需要建立更快的响应机制
  • 供应链安全面临新威胁,攻击者能够伪造代码审查和合并记录,传统的安全控制措施如代码审查和审计日志已不足以防范此类攻击
  • 组织应立即检查Web日志中的"@gl_introduced"请求模式,对未修补的系统实施临时缓解措施,并重新评估供应链安全策略

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 Open Source 开源