MEDIUM6.9

PYSEC-2026-3716

Quick fix

PYSEC-2026-3716 — datasets: upgrade to the fixed version with the command below.

pip install --upgrade 'datasets>=5.0.1'

Details

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.

Are you affected?

Enter the version of the package you're using.

Affected packages

PyPI/datasets

Introduced in: 0Fixed in: 5.0.1

Fixpip install --upgrade 'datasets>=5.0.1'

References

https://github.com/huggingface/datasets/issues/8324[REPORT]
https://github.com/huggingface/datasets/commit/f989ef9b4cc6c0039a7a82458eebca49e2b58b4b[FIX]
https://github.com/huggingface/datasets/pull/8325[FIX]
https://www.vulncheck.com/advisories/datasets-path-traversal-via-unsanitized-file-name-metadata[FIX]
https://github.com/advisories/GHSA-379c-qx7v-6h59[ADVISORY]