CVE-2025-51480
Published: 22 July 2025
Summary
CVE-2025-51480 is a high-severity Path Traversal (CWE-22) vulnerability in Linuxfoundation Onnx. Its CVSS base score is 8.8 (High).
Operationally, exploitation aligns with the MITRE ATT&CK technique Data Destruction (T1485); ranked in the top 41.0% of CVEs by exploit likelihood; it is not currently listed in the CISA KEV catalog; a public proof-of-concept is referenced.
This vulnerability is AI-related — categorised as Deep Learning Frameworks; in the Supply Chain and Deployment risk domain.
The strongest mitigations our analysis identified are NIST 800-53 AC-3 (Access Enforcement) and SI-10 (Information Input Validation).
Deeper analysis
CVE-2025-51480 is a path traversal vulnerability affecting ONNX version 1.17.0, specifically in the onnx.external_data_helper.save_external_data function. Attackers can exploit it by supplying crafted external_data.location paths containing traversal sequences, which bypass intended directory restrictions and enable overwriting of arbitrary files. The issue carries a CVSS v3.1 base score of 8.8 (AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H) and maps to CWE-22; it was published on 2025-07-22.
The vulnerability can be exploited by remote attackers requiring no privileges, but it demands user interaction, such as convincing a victim to load or process a malicious ONNX model file. Successful exploitation allows arbitrary file overwrites on the target system, resulting in high impacts to confidentiality, integrity, and availability, depending on the overwritten files' locations and privileges of the affected process.
Advisories and patches are detailed in the GitHub security advisory GHSA-6rq9-53c3-f7vj, the ONNX repository, and pull requests #6959 and #7040, which address the issue. Further analysis appears in a Gecko Security blog post at https://www.gecko.security/blog/cve-2025-51480. Security practitioners should apply these updates to mitigate risks in ONNX deployments.
ONNX serves as an open format for machine learning model interoperability, heightening the vulnerability's relevance to AI/ML pipelines involving external data handling. No details on real-world exploitation are available in the provided information.
EU & UK References
- 🇪🇺 ENISA EUVD: EUVD-2025-22338
Vulnerability details
Path Traversal vulnerability in onnx.external_data_helper.save_external_data in ONNX 1.17.0 allows attackers to overwrite arbitrary files by supplying crafted external_data.location paths containing traversal sequences, bypassing intended directory restrictions.
- CWE(s)
AI Security AnalysisAI
- AI Category
- Deep Learning Frameworks
- Risk Domain
- Supply Chain and Deployment
- OWASP Top 10 for LLMs 2025
- None mapped
- Classification Reason
- Matched keywords: onnx
Related Threats
MITRE ATT&CK Enterprise TechniquesAI
Why these techniques?
Path traversal enables direct arbitrary file overwrite, facilitating data destruction, encryption for impact, and stored data manipulation.
CVEs Like This One
Affected Assets
Mitigating Controls
Mitigating Controls (NIST 800-53 r5) AI
Directly mitigates the path traversal vulnerability by identifying, reporting, and applying available patches for ONNX 1.17.0 as detailed in GHSA-6rq9-53c3-f7vj and related pull requests.
Requires validation of external_data.location paths from untrusted ONNX model inputs to block traversal sequences like '../' that bypass directory restrictions.
Enforces file system access controls to restrict the ONNX process from writing to arbitrary locations outside intended directories.