MLflow Model Registry Webhook服务端请求伪造漏洞

详细信息

漏洞类型

其他

漏洞描述

MLflow是一个面向代理、大型语言模型和机器学习模型的开源AI工程平台。 MLflow 3.15.0之前版本Model Registry Webhook功能存在服务端请求伪造漏洞，未经身份验证的POST /api/2.0/mlflow/webhooks/{id}/test端点仅对原始URL调用mlflow/utils/validation.py中的_validate_webhook_url()，而mlflow/webhooks/delivery.py会跟随重定向并重新解析主机名而不固定验证过的地址，允许攻击者访问内部或云元数据服务并获取response_status和 response_body。

修复建议

请升级Mlfow到3.15.0或更高版本，参考链接： https://github.com/mlflow/mlflow/security/advisories/GHSA-7gwp-5pfp-969j https://github.com/mlflow/mlflow/issues/24179 https://github.com/mlflow/mlflow/pull/24258 https://github.com/mlflow/mlflow/commit/ba949522477cbd5915aa55d29b0cfad7d5ddf939 https://github.com/mlflow/mlflow/releases/tag/v3.15.0

缓解措施

考虑为MLflow跟踪服务器实现身份验证，或将服务器访问限制在受信任的网络中。避免在生产环境中使用默认的SQLite 后端，以减少风险暴露。

参考链接

• https://github.com/advisories/GHSA-7gwp-5pfp-969j

  GitHub

• https://github.com/mlflow/mlflow/commit/ba949522477cbd5915aa55d29b0cfad7d5ddf939

  CVE · x_refsource_MISC

• https://github.com/mlflow/mlflow/issues/24179

  CVE · x_refsource_MISC

• https://github.com/mlflow/mlflow/pull/24258

  CVE · x_refsource_MISC

• https://github.com/mlflow/mlflow/releases/tag/v3.15.0

  CVE · x_refsource_MISC

• https://github.com/mlflow/mlflow/security/advisories/GHSA-7gwp-5pfp-969j

  CVE · x_refsource_CONFIRM

• https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-64849

  NVD · US Government Resource