Skip to content
OpenAdvisories
High · 8.8 GHSA-gqvg-gmmx-x4hm

MLFLOW_ALLOW_PICKLE_DESERIALIZATION=False safety control bypassed by mlflow.statsmodels flavor — RCE via crafted model artifact

Published Sep 1, 2026 · Updated Sep 1, 2026 · Assigned by GHSA

Description

## Summary

MLflow introduced `MLFLOW_ALLOW_PICKLE_DESERIALIZATION` as a security control to prevent unsafe `pickle.load` execution during model loading, in response to CVE-2024-37052 through CVE-2024-37060. When set to `False`, operators expect all pickle deserialization to be blocked. The most recent related fix (#21188) patched a bypass in the pyfunc flavor.

However, the `mlflow.statsmodels` flavor completely omits this guard. An attacker who places a crafted MLmodel artifact into any accessible artifact store can trigger arbitrary code execution on any process that calls `mlflow.pyfunc.load_model()` against the malicious model — **even when `MLFLOW_ALLOW_PICKLE_DESERIALIZATION=False`**.

This is a security control bypass. The operator believes pickle RCE is mitigated; the statsmodels flavor silently ignores the control.

---

## Root Cause

`mlflow.pyfunc.load_model()` dispatches to flavor `_load_pyfunc` implementations via:

```
# mlflow/pyfunc/__init__.py L1170-1172
model_impl = importlib.import_module(conf[MAIN])._load_pyfunc(data_path)
```

The guarded pattern (from `mlflow/sklearn/__init__.py` L526-533, the reference implementation) is:

```
if (
not MLFLOW_ALLOW_PICKLE_DESERIALIZATION.get()
and not is_in_databricks_runtime()
and not is_in_databricks_model_serving_environment()
):
raise MlflowException("Deserializing model using pickle is disallowed...")
```

`mlflow/statsmodels/__init__.py` has **no such check**:

```
# L307-320 — no guard anywhere in this file
def _load_model(path):
import statsmodels.iolib.api as smio
return smio.load_pickle(path) # calls pickle.load() directly

def _load_pyfunc(path):
return _StatsmodelsModelWrapper(_load_model(path))
```

`statsmodels.iolib.api.load_pickle` is a thin wrapper around `pickle.load`. Its own docstring warns: *"Never unpickle data received from an untrusted or unauthenticated source."*

---

## Trigger

An attacker crafts an MLmodel YAML that specifies `mlflow.statsmodels` as the loader module:

```
flavors:
python_function:
loader_module: mlflow.statsmodels
data: model.pkl
statsmodels:
data: model.pkl
statsmodels_version: 0.14.0
```

With a malicious `model.pkl` placed alongside it in the artifact store, any call to:

```
os.environ["MLFLOW_ALLOW_PICKLE_DESERIALIZATION"] = "False"
mlflow.pyfunc.load_model("models:/MaliciousModel/1")
```

...deserializes the pickle file with **no guard check**, executing arbitrary code with the privileges of the calling process.

On default MLflow deployments (no `--app-name basic-auth`), authentication is disabled, so artifact upload requires no credentials.

---

## Affected Code

- `mlflow/statsmodels/__init__.py` L307-310: `_load_model` — calls `smio.load_pickle` without checking `MLFLOW_ALLOW_PICKLE_DESERIALIZATION`
- `mlflow/statsmodels/__init__.py` L313-320: `_load_pyfunc` — dispatches to `_load_model` without checking the control

Permalink (commit `0b0c576c`):
- https://github.com/mlflow/mlflow/blob/0b0c576c642b5b0d9496c829809c7d097403bc9f/mlflow/statsmodels/__init__.py#L307-L310
- https://github.com/mlflow/mlflow/blob/0b0c576c642b5b0d9496c829809c7d097403bc9f/mlflow/statsmodels/__init__.py#L313-L320

---

## Recommended Fix

Add the missing guard to `mlflow/statsmodels/__init__.py`:

```
from mlflow.environment_variables import MLFLOW_ALLOW_PICKLE_DESERIALIZATION
from mlflow.utils.databricks_utils import (
is_in_databricks_model_serving_environment,
is_in_databricks_runtime,
)

def _load_model(path):
if (
not MLFLOW_ALLOW_PICKLE_DESERIALIZATION.get()
and not is_in_databricks_runtime()
and not is_in_databricks_model_serving_environment()
):
raise MlflowException(
"Deserializing model using pickle is disallowed, but this statsmodels "
"model requires pickle deserialization. Set environment variable "
"'MLFLOW_ALLOW_PICKLE_DESERIALIZATION' to 'true' to allow this."
)
import statsmodels.iolib.api as smio
return smio.load_pickle(path)
```

Affected packages

EcosystemPackageAffectedFixedRecord
PyPI mlflow >= 2.1.0, < 3.15.0 3.15.0 GHSA-gqvg-gmmx-x4hm

References