Skip to main content

Onnx

9 CVEs product

Monthly

CVE-2026-34445 Apr 01, 17:30 PyPI HIGH PATCH GHSA This Week

Arbitrary attribute injection in ONNX Python library (versions prior to 1.21.0) allows unauthenticated remote attackers to manipulate internal object properties by embedding malicious metadata in ONNX model files, resulting in potential information disclosure, data integrity violations, and high availability impact (CVSS 8.6). The vulnerability stems from unchecked use of Python's setattr() with externally-controlled keys during ExternalDataInfo deserialization. No public exploit code or CISA KEV listing identified at time of analysis, but proof-of-concept development is trivial given the straightforward nature of Python attribute manipulation. EPSS data not provided, but the unauthenticated network-accessible attack vector and low complexity suggest material risk for organizations processing untrusted ONNX models.

Python Information Disclosure Microsoft Red Hat Suse +1
NVD GitHub
CVSS 3.1
8.6
EPSS
0.0%
CVE-2026-14647 Jul 04, 19:00 LOW POC PATCH Monitor

Out-of-bounds read in ONNX versions up to 1.21.x exposes limited memory contents to low-privileged remote attackers via the convPoolShapeInference_opset19 shape inference function. The CVSS 4.0 score of 2.1 reflects minimal real-world impact - confidentiality-only, low severity - yet a public proof-of-concept is available via GitHub issue #8036. No active exploitation has been confirmed by CISA KEV, and an upstream patch exists at commit a7bf3a0f1d18bb62575236ef6e4944980c40e045 via PR #8051.

Buffer Overflow Information Disclosure N A Onnx
NVD VulDB GitHub
CVSS 4.0
2.1
EPSS
0.3%
CVE-2026-11329 Jun 05, 12:15 LOW PATCH Monitor

Weak cache key construction in onnx-mlir's torch backend (versions up to 0.5.0.0) omits tensor data type (dtype) from placeholder node hash keys, enabling cache collisions between semantically distinct nodes. A locally authenticated attacker with high-complexity manipulation can cause the compiler to incorrectly reuse cached compilation results across mismatched dtypes, yielding low-integrity and low-availability impacts. No public exploit is identified at time of analysis; the upstream fix is confirmed via commit 72c5187 and PR #3427.

Python Information Disclosure Onnx Onnx Mlir
NVD VulDB GitHub
CVSS 4.0
2.0
EPSS
0.0%
CVE-2026-34447 Apr 01, 17:39 PyPI MEDIUM PATCH GHSA This Month

ONNX versions prior to 1.21.0 allow local attackers to read arbitrary files outside the model directory through symlink traversal during external data loading, requiring user interaction to load a malicious model file. The vulnerability has a CVSS score of 5.5 (medium severity) and is classified as information disclosure with confirmed patch availability in version 1.21.0.

Information Disclosure Microsoft Suse Onnx
NVD GitHub
CVSS 3.1
5.5
EPSS
0.0%
CVE-2026-34446 Apr 01, 17:37 PyPI MEDIUM PATCH GHSA This Month

ONNX versions prior to 1.21.0 allow local attackers to read arbitrary files by exploiting a hardlink-based path traversal vulnerability in onnx.load(). The vulnerability bypasses existing symlink protections because hardlinks appear as regular files to filesystem checks. An attacker with local file system access can craft a malicious ONNX model file using hardlinks to access sensitive data outside the intended directory, requiring user interaction to load the crafted model. No public exploit code has been identified; EPSS score of 4.7 indicates low exploitation probability despite moderate CVSS impact.

Path Traversal Microsoft Red Hat Suse Onnx
NVD GitHub
CVSS 3.1
4.7
EPSS
0.0%
CVE-2024-7776 Mar 20, 10:15 PyPI CRITICAL POC PATCH Act Now

A vulnerability in the `download_model` function of the onnx/onnx framework, before and including version 1.16.1, allows for arbitrary file overwrite due to inadequate prevention of path traversal. Rated critical severity (CVSS 9.1), this vulnerability is remotely exploitable, no authentication required, low attack complexity. Public exploit code available and no vendor patch available.

Path Traversal Suse Onnx
NVD
CVSS 3.1
9.1
EPSS
1.5%
CVE-2024-5187 Jun 06, 19:16 PyPI HIGH POC PATCH This Week

A vulnerability in the `download_model_with_test_data` function of the onnx/onnx framework, version 1.16.0, allows for arbitrary file overwrite due to inadequate prevention of path traversal attacks. Rated high severity (CVSS 8.8), this vulnerability is remotely exploitable, no authentication required, low attack complexity. Public exploit code available and no vendor patch available.

RCE Path Traversal Onnx Linuxfoundation
NVD
CVSS 3.1
8.8
EPSS
1.2%
CVE-2024-27319 Feb 23, 18:15 PyPI CRITICAL PATCH Act Now

Versions of the package onnx before and including 1.15.0 are vulnerable to Out-of-bounds Read as the ONNX_ASSERT and ONNX_ASSERTM functions have an off by one string copy. Rated critical severity (CVSS 9.1), this vulnerability is remotely exploitable, no authentication required, low attack complexity.

Buffer Overflow Information Disclosure Fedora Onnx Linuxfoundation +1
NVD GitHub
CVSS 3.1
9.1
EPSS
0.6%
CVE-2024-27318 Feb 23, 18:15 PyPI HIGH PATCH This Week

Versions of the package onnx before and including 1.15.0 are vulnerable to Directory Traversal as the external_data field of the tensor proto can have a path to the file which is outside the model. Rated high severity (CVSS 7.5), this vulnerability is remotely exploitable, no authentication required, low attack complexity. This Path Traversal vulnerability could allow attackers to access files and directories outside the intended path.

Path Traversal Fedora Onnx Linuxfoundation Fedoraproject
NVD GitHub
CVSS 3.1
7.5
EPSS
1.2%
EPSS 0% CVSS 8.6
HIGH PATCH This Week

Arbitrary attribute injection in ONNX Python library (versions prior to 1.21.0) allows unauthenticated remote attackers to manipulate internal object properties by embedding malicious metadata in ONNX model files, resulting in potential information disclosure, data integrity violations, and high availability impact (CVSS 8.6). The vulnerability stems from unchecked use of Python's setattr() with externally-controlled keys during ExternalDataInfo deserialization. No public exploit code or CISA KEV listing identified at time of analysis, but proof-of-concept development is trivial given the straightforward nature of Python attribute manipulation. EPSS data not provided, but the unauthenticated network-accessible attack vector and low complexity suggest material risk for organizations processing untrusted ONNX models.

Python Information Disclosure Microsoft +3
NVD GitHub
EPSS 0% CVSS 2.1
LOW POC PATCH Monitor

Out-of-bounds read in ONNX versions up to 1.21.x exposes limited memory contents to low-privileged remote attackers via the convPoolShapeInference_opset19 shape inference function. The CVSS 4.0 score of 2.1 reflects minimal real-world impact - confidentiality-only, low severity - yet a public proof-of-concept is available via GitHub issue #8036. No active exploitation has been confirmed by CISA KEV, and an upstream patch exists at commit a7bf3a0f1d18bb62575236ef6e4944980c40e045 via PR #8051.

Buffer Overflow Information Disclosure N A +1
NVD VulDB GitHub
EPSS 0% CVSS 2.0
LOW PATCH Monitor

Weak cache key construction in onnx-mlir's torch backend (versions up to 0.5.0.0) omits tensor data type (dtype) from placeholder node hash keys, enabling cache collisions between semantically distinct nodes. A locally authenticated attacker with high-complexity manipulation can cause the compiler to incorrectly reuse cached compilation results across mismatched dtypes, yielding low-integrity and low-availability impacts. No public exploit is identified at time of analysis; the upstream fix is confirmed via commit 72c5187 and PR #3427.

Python Information Disclosure Onnx +1
NVD VulDB GitHub
EPSS 0% CVSS 5.5
MEDIUM PATCH This Month

ONNX versions prior to 1.21.0 allow local attackers to read arbitrary files outside the model directory through symlink traversal during external data loading, requiring user interaction to load a malicious model file. The vulnerability has a CVSS score of 5.5 (medium severity) and is classified as information disclosure with confirmed patch availability in version 1.21.0.

Information Disclosure Microsoft Suse +1
NVD GitHub
EPSS 0% CVSS 4.7
MEDIUM PATCH This Month

ONNX versions prior to 1.21.0 allow local attackers to read arbitrary files by exploiting a hardlink-based path traversal vulnerability in onnx.load(). The vulnerability bypasses existing symlink protections because hardlinks appear as regular files to filesystem checks. An attacker with local file system access can craft a malicious ONNX model file using hardlinks to access sensitive data outside the intended directory, requiring user interaction to load the crafted model. No public exploit code has been identified; EPSS score of 4.7 indicates low exploitation probability despite moderate CVSS impact.

Path Traversal Microsoft Red Hat +2
NVD GitHub
EPSS 1% CVSS 9.1
CRITICAL POC PATCH Act Now

A vulnerability in the `download_model` function of the onnx/onnx framework, before and including version 1.16.1, allows for arbitrary file overwrite due to inadequate prevention of path traversal. Rated critical severity (CVSS 9.1), this vulnerability is remotely exploitable, no authentication required, low attack complexity. Public exploit code available and no vendor patch available.

Path Traversal Suse Onnx
NVD
EPSS 1% CVSS 8.8
HIGH POC PATCH This Week

A vulnerability in the `download_model_with_test_data` function of the onnx/onnx framework, version 1.16.0, allows for arbitrary file overwrite due to inadequate prevention of path traversal attacks. Rated high severity (CVSS 8.8), this vulnerability is remotely exploitable, no authentication required, low attack complexity. Public exploit code available and no vendor patch available.

RCE Path Traversal Onnx +1
NVD
EPSS 1% CVSS 9.1
CRITICAL PATCH Act Now

Versions of the package onnx before and including 1.15.0 are vulnerable to Out-of-bounds Read as the ONNX_ASSERT and ONNX_ASSERTM functions have an off by one string copy. Rated critical severity (CVSS 9.1), this vulnerability is remotely exploitable, no authentication required, low attack complexity.

Buffer Overflow Information Disclosure Fedora +3
NVD GitHub
EPSS 1% CVSS 7.5
HIGH PATCH This Week

Versions of the package onnx before and including 1.15.0 are vulnerable to Directory Traversal as the external_data field of the tensor proto can have a path to the file which is outside the model. Rated high severity (CVSS 7.5), this vulnerability is remotely exploitable, no authentication required, low attack complexity. This Path Traversal vulnerability could allow attackers to access files and directories outside the intended path.

Path Traversal Fedora Onnx +2
NVD GitHub

This site uses cookies essential for authentication and security. No tracking or analytics cookies are used. Privacy Policy