cve.bz

CVE-2026-49114 linuxfoundation onnx: Path traversal

6.8 mediumExploit referenceEPSS 0.2%Fix available

Medium (6.8) in linuxfoundation onnx. EPSS 0.2% estimated probability of exploitation activity in the next 30 days. Fixed in onnx 1.21.0.

On this page
In ONNX before 1.21.0, the 'save_external_data' function builds the external-data file path from the model's external_data location field and opens it for writing without 'O_NOFOLLOW/O_EXCL', after a non-atomic 'os.path.isfile()' check. A local attacker with write access to the directory where a victim serializes external data can deterministically pre-plant a symlink that is being followed, causing the victim's write to append to any file the victim can write, e.g. ~/.ssh/authorized_keys, cron files, or application configs. Fixed in 1.21.0.

Severity

CVSS 4.0

6.8 medium
CVSS:4.0/AV:L/AC:L/AT:N/PR:N/UI:A/VC:N/VI:H/VA:H/SC:N/SI:N/SA:N

CVSS 3.1

7.1 high
CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:N/I:H/A:H
EPSS
0.2% (unchanged from 0.2% a week ago; 0.2% 30 days ago) estimated probability of exploitation activity in the next 30 days; 5th percentile. Scored 2026-11-17. A forecast for this CVE across the internet, not for your own systems.
CISA's assessment (SSVC)
Exploitation: PoC · Automatable: no · Technical impact: partial, assessed 2026-08-21. From CISA's Vulnrichment in the CVE List: whether it is exploited, whether exploiting it can be automated, and how much control it gives.
CWEs
CWE-22 Improper Limitation of a Pathname to a Restricted Directory (Path traversal)
CWE-59 Improper Link Resolution Before File Access (Link Following)
CWE-367 Time-of-check Time-of-use (TOCTOU) Race Condition
OWASP Top 10
A01:2025 Broken Access Control
Published
Modified
NVD analysis
Analyzed
Record from
the NVD, synced

Fix and patched versions

Ask an AI about this CVE

Show the promptExplain CVE-2026-49114 for an engineer deciding whether to act. Use https://cve.bz/api/public/cves/CVE-2026-49114?via=ask-ai-copy and https://cve.bz/cve/CVE-2026-49114?via=ask-ai-copy as the source of truth: what is affected and which versions fix it, whether it is in CISA KEV, public exploits, and its EPSS (an internet-wide 30-day estimate of exploitation activity, not a forecast for my systems). If they give no fixed versions, use the CVE record https://cveawg.mitre.org/api/cve/CVE-2026-49114 or the vendor advisory, not third-party mirrors. Tell me how to check whether I am affected and how to verify a fix. Cite the sources and say where they are uncertain.

Opens a third-party assistant; its answer is its own, not cve.bz's. The prompt holds only this CVE's id and public links.

Affected

Versions

From the CNA's record in the CVE List.

ProductAffected
ONNXbefore 1.21.0

Packages (OSV)

AdvisoryEcosystemPackageAffectedFixed in
GHSA-q56x-g2fj-4rj6PyPIonnx<1.21.01.21.0
PYSEC-2026-3876PyPIonnx<1.21.01.21.0

Linux distributions

DistributionReleasePackageFixed inAdvisory
Debian12, 13, 14onnxno fix listedDEBIAN-CVE-2026-49114
Ubuntu22.04:LTS, 26.04:LTS, 26.10 and 1 moreonnxno fix listedUBUNTU-CVE-2026-49114

From the distributions' data via OSV.dev: the Debian Security Tracker; Ubuntu Security Notices © Canonical Ltd., CC BY-SA 4.0. Modified: reformatted and linked to CVE records. Credits. On the distributions' own pages: Debian · Ubuntu.

Red Hat

Red Hat rates this Important.

Red Hat's statement

An IMPORTANT severity flaw was found in ONNX, affecting Red Hat OpenShift AI (RHOAI) components that ship a vulnerable onnx version. The save_external_data function builds the external-data file path from the model's external_data location field and opens it for writing without symlink protections (O_NOFOLLOW/O_EXCL), after a non-atomic os.path.isfile() check. A local attacker with write access to a directory in which a victim serializes ONNX external data can exploit this time-of-check-to-time-of-use (TOCTOU) race condition by pre-planting a symlink that is followed when the victim saves the model, causing the victim's write to append to arbitrary files the victim can write, such as ~/.ssh/authorized_keys, cron files, or application configuration files. Successful exploitation can result in unauthorized modification of critical files and further compromise of the victim's account. Exploitation requires local access, write access to the serialization directory, and that the victim save a model with external data; the attack vector is therefore local, not remote. Within RHOAI, the components that ship a vulnerable onnx version are affected, and a fix is being pursued through the responsible delegated product team; the remaining evaluated RHOAI components are not affected because they ship an onnx version that already contains the fix. Red Hat has rated this flaw as having a security impact of IMPORTANT.

Red Hat's mitigation

Mitigation for this issue is either not available or the currently available options do not meet the Red Hat Product Security criteria comprising ease of use and deployment, applicability to widespread installation base, or stability. The upstream project fixed this issue in onnx 1.21.0, so upgrading to a release that includes this fix is recommended.

ProductPackageStateAdvisory
Red Hat OpenShift AI 2.25rhoai/odh-openvino-model-server-rhel9:1791275700Fixed 2026-10-21RHSA-2026:87410
Red Hat OpenShift AI 3.4rhoai/odh-pipeline-runtime-datascience-cpu-py312-rhel9:1791829804Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-pipeline-runtime-pytorch-cuda-py312-rhel9:1791829182Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-pipeline-runtime-pytorch-llmcompressor-cuda-py312-rhel9:1791829658Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-pipeline-runtime-pytorch-rocm-py312-rhel9:1791868244Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-pipeline-runtime-tensorflow-cuda-py312-rhel9:1791828837Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-pipeline-runtime-tensorflow-rocm-py312-rhel9:1791868241Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-workbench-codeserver-datascience-cpu-py312-rhel9:1791788267Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-workbench-jupyter-datascience-cpu-py312-rhel9:1791872084Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-workbench-jupyter-pytorch-cuda-py312-rhel9:1791829615Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-workbench-jupyter-pytorch-llmcompressor-cuda-py312-rhel9:1791832538Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-workbench-jupyter-pytorch-rocm-py312-rhel9:1791868194Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-workbench-jupyter-tensorflow-cuda-py312-rhel9:1791833225Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-workbench-jupyter-tensorflow-rocm-py312-rhel9:1791868238Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI 3.4rhoai/odh-workbench-jupyter-trustyai-cpu-py312-rhel9:1791829530Fixed 2026-10-19RHSA-2026:84734
Red Hat OpenShift AI (RHOAI)rhoai/odh-mlserver-cuda-rhel9Not affected
Red Hat OpenShift AI (RHOAI)rhoai/odh-mlserver-rhel9Not affected

Contains Red Hat Security Data © Red Hat, Inc., licensed under CC BY 4.0. Source: https://access.redhat.com/security/data. Modified: reformatted. Red Hat's page for CVE-2026-49114.

Attack patterns (CAPEC)

  • CAPEC-27 Leveraging Race Conditions via Symbolic Links Likelihood: Medium · Severity: High
  • CAPEC-29 Leveraging Time-of-Check and Time-of-Use (TOCTOU) Race Conditions Likelihood: High · Severity: High
  • CAPEC-17 Using Malicious Files Likelihood: High · Severity: Very High
  • CAPEC-35 Leverage Executable Code in Non-Executable Files Likelihood: High · Severity: Very High
  • CAPEC-76 Manipulating Web Input to File System Calls Likelihood: High · Severity: Very High
  • CAPEC-132 Symlink Attack Likelihood: Low · Severity: High
  • CAPEC-64 Using Slashes and URL Encoding Combined to Bypass Validation Logic Likelihood: High · Severity: High
  • CAPEC-78 Using Escaped Slashes in Alternate Encoding Likelihood: High · Severity: High
  • CAPEC-79 Using Slashes in Alternate Encoding Likelihood: High · Severity: High
  • CAPEC-126 Path Traversal Likelihood: High · Severity: Very High

The attack patterns MITRE's CAPEC relates to this record's weaknesses, as MITRE names them. Credits.

References

Check your stackJSON