Datasets Path Traversal via Unsanitized file_name Metadata
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.
CVE-2026-66007
Primary source · Revision 1
Imported from structured advisory data. Source references identify upstream records or fields; verify affected versions and your deployed configuration.
Security context
Source reference: cve:CVE-2026-66007
datasets
Source reference: /containers/cna/affected/0/product
- Affected versions
- Not stated in this record
- Fixed versions
- Not stated in this record
- Prerequisites
- Not stated in this record
Structured affected-version data
Original package identities and version rules from this source.
[
{
"repo": "https://github.com/huggingface/datasets",
"vendor": "huggingface",
"product": "datasets",
"versions": [
{
"status": "affected",
"version": "0",
"versionType": "semver",
"lessThanOrEqual": "5.0.0"
},
{
"status": "unaffected",
"version": "f989ef9b4cc6c0039a7a82458eebca49e2b58b4b",
"versionType": "git"
}
],
"packageURL": "pkg:pypi/datasets",
"defaultStatus": "unaffected"
}
]- Related CVEs
- CVE-2026-66007
Source reference:
/cveMetadata/cveId
Publisher severity
medium
Source reference: /containers/cna/metrics
- Event
- #5449 · Revision 1
- Added to Heist Feed
- Publisher date