MLflow's statsmodel flavor, versions 2.1.0 to 3.14.0, omits the MLFLOW_ALLOW_PICKLE_DESERIALIZATION=False security control entirely in _load_model(), which allows a remote attacker to execute arbitrary code via a crafted MLmodel artifact.
References
| Link | Resource |
|---|---|
| https://kb.cert.org/vuls/id/369093 |
Configurations
No configuration.
History
23 Sep 2026, 17:17
| Type | Values Removed | Values Added |
|---|---|---|
| New CVE |
Information
Published : 2026-09-23 17:17
Updated : 2026-09-24 04:18
NVD link : CVE-2026-96804
Mitre link : CVE-2026-96804
CVE.ORG link : CVE-2026-96804
JSON object : View
Products Affected
No product.
CWE
CWE-502
Deserialization of Untrusted Data
