Actively exploited KEV catalog No authentication
CVE-2026-64849

MLflow Server-Side Request Forgery Vulnerability

MLflow is an open source AI engineering platform for agents, large language models, and machine learning models. Starting in 3.3.0 and prior to 3.15.0, the unauthenticated POST /api/2.0/mlflow/webhooks/{id}/test endpoint calls…

CheckOps priority index
6.4/10 Patch within 7 days
Severity9.3
Probability1.0
Exploitation4/5

Severity = CVSS. Probability = EPSS (FIRST). Exploitation = public signals: KEV, EUVD, ransomware, EPSS above 10 %, no account required.

EPSS · today 10.0% higher than 95.6% of CVEs
CVSS 3.1 9.3 AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:
Exploited for 76 days 19 Aug 2026
CISA due date 02 Sept 2026 for US federal agencies

Exploitation probability EPSS · FIRST

05010025 Sep05 Oct15 Oct24 Oct02 Nov
EPSS probability (%)Percentile among all CVEs

Exploitation signals 4 / 5

  • KEV catalog (CISA)19 Aug 2026
  • Exploited in EUVD (ENISA)19 Aug 2026
  • Used by ransomwareno
  • EPSS above 10 %10.0%
  • No account, over the networkyes

What to do

Fix

Apply the vendor’s fix as soon as it is available.

CISA guidance: Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA’s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA’s “Forensics Triage Requirements” (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset's internet exposure and ensuring adherence to BOD 26-04 patching guidelines.

Meanwhile

Take the service off the internet or restrict access to it, watch the logs and look for signs of compromise.

Manufacturers (CRA)

Exploited vulnerability in your product: early warning to ENISA within 24 h, then notification within 72 h (Article 14 of the Cyber Resilience Act). The CRA explained

Description ENISA EUVD

MLflow is an open source AI engineering platform for agents, large language models, and machine learning models. Starting in 3.3.0 and 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.

Affected products

MLflow
Affected versions
  • 3.3.0, < 3.15.0
Weaknesses (CWE)

Open source packages OSV

Affected libraries according to OSV, with fixed versions.

Timeline

  1. Published

  2. Exploitation reported (ENISA)

  3. Added to the KEV catalog

  4. CISA remediation due date

  5. EPSS peak: 10.0%

References

Are you affected?

CheckOps reads your SBOMs and scans, and alerts you as soon as one of your dependencies is exploited.

Get notified at launch

Sources: CISA KEV · ENISA EUVD · FIRST EPSS · OSV. Public data aggregated by CheckOps, updated automatically. Source texts remain the property of their authors. All exploited vulnerabilities