Описание
MLflow allows unauthorized access to multipart upload endpoints when the --serve-artifacts mode is enabled
A vulnerability in MLflow versions <=3.10.1.dev0 allows unauthorized access to multipart upload (MPU) endpoints when the --serve-artifacts mode is enabled. The authorization logic does not enforce resource-level permission checks for /mlflow-artifacts/mpu/* endpoints, enabling attackers to overwrite artifacts belonging to other users. This can lead to unauthorized cross-user writes, model supply chain poisoning, and arbitrary code execution when compromised models are loaded. The issue is resolved in version 3.10.0.
Ссылки
- https://nvd.nist.gov/vuln/detail/CVE-2026-2651
- https://github.com/mlflow/mlflow/commit/d7290811d8f3c95366d80109424edc1fb1ad966f
- https://access.redhat.com/security/cve/CVE-2026-2651
- https://bugzilla.redhat.com/show_bug.cgi?id=2481117
- https://huntr.com/bounties/65beb119-d3e0-4e03-af2f-fa98f78f83dc
- https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-2651.json
Пакеты
mlflow
< 3.11.0rc0
3.11.0rc1
Связанные уязвимости
A vulnerability in MLflow versions <=3.10.1.dev0 allows unauthorized access to multipart upload (MPU) endpoints when the `--serve-artifacts` mode is enabled. The authorization logic does not enforce resource-level permission checks for `/mlflow-artifacts/mpu/*` endpoints, enabling attackers to overwrite artifacts belonging to other users. This can lead to unauthorized cross-user writes, model supply chain poisoning, and arbitrary code execution when compromised models are loaded. The issue is resolved in version 3.10.0.
A vulnerability in MLflow versions <=3.10.1.dev0 allows unauthorized access to multipart upload (MPU) endpoints when the `--serve-artifacts` mode is enabled. The authorization logic does not enforce resource-level permission checks for `/mlflow-artifacts/mpu/*` endpoints, enabling attackers to overwrite artifacts belonging to other users. This can lead to unauthorized cross-user writes, model supply chain poisoning, and arbitrary code execution when compromised models are loaded. The issue is resolved in version 3.10.0.
Уязвимость платформы управления жизненным циклом моделей машинного обучения MLflow, связанная с недостатками процедуры авторизации, позволяющая нарушителю выполнить произвольный код