datasetsGHSA-379c-qx7v-6h59
Hugging Face Datasets folder-based builders allow path traversal through file_name metadata
Medium6.5CVE-2026-66007 · Published Jul 24, 2026 · updated Oct 2, 2026
Datasets through 5.0.0, fixed in 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.
Affected versions
| Package | Affected | Fixed in |
|---|---|---|
| datasets PyPI | < 5.0.1 | 5.0.1 |
Details and references
- CVSS 3.1
- CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:N/A:N
- Severity from
- GitHub (reviewed advisory)
- Weakness
- CWE-22
- Also known as
- CVE-2026-66007, PYSEC-2026-3716
- nvd.nist.gov/vuln/detail/CVE-2026-66007
- github.com/huggingface/datasets/issues/8324
- github.com/huggingface/datasets/pull/8325
- github.com/huggingface/datasets/commit/f989ef9b4cc6c0039a7a82458eebca49e2b58b4b
- github.com/huggingface/datasets
- github.com/pypa/advisory-database/tree/main/vulns/datasets/PYSEC-2026-3716.yaml
- www.vulncheck.com/advisories/datasets-path-traversal-via-unsanitized-file-name-metadata
More datasets advisories
All datasets| Date | Advisory | Severity | Fixed in |
|---|---|---|---|
| Jul 24 | datasets: path traversal Medium6.5Jul 24 | Medium6.5 | 5.0.1 |