Vulnerability Description
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.
CVSS Score
MEDIUM
Affected Products
| Vendor | Product | Versions |
|---|---|---|
| Huggingface | Datasets | <= 5.0.0 |
Related Weaknesses (CWE)
References
- https://github.com/huggingface/datasets/commit/f989ef9b4cc6c0039a7a82458eebca49ePatch
- https://github.com/huggingface/datasets/issues/8324ExploitIssue TrackingVendor Advisory
- https://github.com/huggingface/datasets/pull/8325Issue TrackingPatch
- https://www.vulncheck.com/advisories/datasets-path-traversal-via-unsanitized-filExploitPatchThird Party Advisory
FAQ
What is CVE-2026-66007?
CVE-2026-66007 is a vulnerability with a CVSS score of 6.5 (MEDIUM). 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 join...
How severe is CVE-2026-66007?
CVE-2026-66007 has been rated MEDIUM with a CVSS base score of 6.5/10. Review the CVSS metrics above for detailed severity breakdown.
Is there a patch for CVE-2026-66007?
Check the references section above for vendor advisories and patch information. Affected products include: Huggingface Datasets.