Path Traversal in Huggingface Datasets
CVE-2026-66007
Datasets through 5.0.0, fixed in commit f989ef9, contains a path traversal vulnerability in folder-based dataset builders where the file_name metadata field is not properly validated before being joined to the dataset directory. Attackers can supply crafted file_name values with directory traversal sequences to read arbitrary local files, which are then embedded into output when save_to_disk or push_to_hub is called.
Vulnerability class: Path Traversal (Directory Traversal)
EPSS: 0.006 (44.4th percentile) — read the EPSS interpretation.
CVSS v3 metric
CVSS v3 base score 6.5 (Medium). Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:N/A:N.
Affected products
- Huggingface Datasets — versions 0, f989ef9b4cc6c0039a7a82458eebca49e2b58b4b
Weakness classification (CWE)
References
- disclosure@vulncheck.com (Patch, patch)
- disclosure@vulncheck.com (issue-tracking, Exploit, technical-description, exploit, Issue Tracking, Vendor Advisory)
- disclosure@vulncheck.com (issue-tracking, Patch, patch, Issue Tracking)
- disclosure@vulncheck.com (Exploit, Patch, Third Party Advisory, third-party-advisory)
Frequently asked questions
- What is CVE-2026-66007?
- CVE-2026-66007 is a medium-severity vulnerability in Huggingface Datasets, classified under Path Traversal. CVSS score: 6.5/10. Published 2026-07-24.
- How severe is CVE-2026-66007?
- Medium severity. CVSS v3 base score is 6.5 out of 10.