CVE-2026-64849
KEVCRITICAL 9.3EPSS 16.4%
MLflow is an open source AI engineering platform for agents, large language models, and machine learning models. Prior to 3.15.0, the unauthenticated POST /api/2.0/mlflow/webhooks/{id}/test endpoint calls _validate_webhook_url() in mlflow/utils/validation.py only for the original URL while mlflow/webhooks/delivery.py follows redirects and re-resolves the hostname without pinning the validated address, allowing attackers to reach internal or cloud metadata services and receive response_status and response_body. This issue is fixed in version 3.15.0.
- CVSS v3.1
- 9.3 CRITICAL
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:N - CVSS v3.1
- 9.3 CRITICAL
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:N - EPSS
- 16.41% chance of exploitation in the next 30 days, 97th percentile
- CISA KEV
- added 2026-08-19
- Nuclei
- critical · CWE-918
- Published
- 2026-08-17
- Updated
- 2026-08-20
Proof-of-concept exploits (5)
- advisories/GHSA-7gwp-5pfp-969j
- BiuTrap/CVE-2026-648490★ · 2026-08-19
- zavisco/CVE-2026-64849.yaml0★ · 2026-08-19
- codeb0ssx/CVE-2026-64849-PoC0★ · 2026-08-18
- mrzero-cool/cve-poc-search0★ · 2026-08-30