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.
Attack Parameters
Attack Vector
Network
Can be exploited remotely
Attack Complexity
Low
Easy to exploit
Attack Requirements
None
No additional conditions
Privileges Required
None
No privileges needed
User Interaction
Active
User action required
Impact Assessment
Confidentiality
High
Complete data leak
Integrity
None
No data modification
Availability
None
No disruption
CVSS Vector v4.0
Weakness Type (CWE)
References 4
https://github.com/huggingface/datasets/commit/f989ef9b4cc6c0039a7a82458eebca49…
disclosure@vulncheck.com
https://github.com/huggingface/datasets/issues/8324
disclosure@vulncheck.com
https://github.com/huggingface/datasets/pull/8325
disclosure@vulncheck.com
https://www.vulncheck.com/advisories/datasets-path-traversal-via-unsanitized-fi…
disclosure@vulncheck.com