CVE-2026-64849
MLflow: Unauthenticated full-read SSRF in webhook delivery: _validate_webhook_url bypassed via unvalidated HTTP redirects (and DNS rebinding)
- Published
- Aug 17, 2026
- Updated
- Aug 20, 2026
- Assigning CNA
- GitHub_M
- Evidence observed
- Aug 19, 2026
Primary CVSS
nvd · CVSS 3.1
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:NModerate · next 30 days
- Percentile
- 96.9%
- Model date
- Sep 21, 2026
EPSS is a statistical estimate, not a certainty or a measure of impact. Combine it with CVSS, KEV status, exposure and your environment.
CISA Known Exploited
This CVE appears in the CISA Known Exploited Vulnerabilities catalog.
Summary
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.
Sources
4- CVE-2026-64849.yamlExploit
Detects unauthenticated MLflow webhook SSRF (CVE-2026-64849) that accesses internal or cloud metadata services and leaks response details via redirect bypass.
Proof-of-concept exploit for CVE-2026-64849: triggers SSRF in MLflow webhook API via crafted POST, fetching cloud instance metadata from 169.254.169.254.
Responsible use
Use vulnerability information only on systems you own or are authorized to test. Kitploit links to public research metadata and does not store exploit code or malicious payloads.