High severity · published Sep 23, 2026
CVE-2026-96804
No description published yet
8.8
CVSS
0.7%
EPSS
No
On KEV
Description
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.
CVSS vector
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
Attack vector
Network
Complexity
Low
Privileges
Low
User interaction
None
Confidentiality
High
Integrity
High
Availability
High
Scope
Unchanged
Exploitation
Is it being used?
Exploit prediction (EPSS)0.7% · higher than 55% of CVEs
Exploited in the wildNo evidence yet
CISA KEVNot listed
Ransomware campaign useNot known
Affected products
Which versions
ML
MLflow · MLflow
2.1 before 3.15.0 · source: vendor
References
NVD
https://nvd.nist.gov/vuln/detail/CVE-2026-96804
CVE.org
https://www.cve.org/CVERecord?id=CVE-2026-96804
kb.cert.org
https://kb.cert.org/vuls/id/369093
More from MLflow
9.3
CVE-2026-64849
MLflow: Unauthenticated full-read SSRF in webhook delivery: _validate_webhook_url bypassed via unvalidated HTTP redirects (and DNS rebinding)
9.8
CVE-2026-116692
MLflow: Remote code execution on unauthenticated MLflow servers via third-party scorer deserialization importing caller-placed artifact modules
8.8
CVE-2026-96775
MLflow dspy bypasses pickle deserialization control
8.8
CVE-2026-13484
MLflow Experiment-scoped Label Schema CRUD API authorization
8.6
CVE-2026-79721
Code execution can occur in versions of the MLflow platform running version 0.0.1 or newer, enabling a maliciously crafted model artifact to execute arbitrary code on an end…