CVE-2026-57845NVD
Vulnerability Summary
Hugging Face Datasets through 5.1.0 contains an insecure deserialization vulnerability that allows attackers to execute arbitrary Python code by crafting a malicious pickle file loaded through the deprecated pandas packaged builder. Attackers can supply a pickle-backed dataset that is processed by Pandas._generate_tables via pandas.read_pickle() without validation, so when a victim calls datasets.load_dataset("pandas", data_files=...), the embedded __reduce__ payload executes code with the victim process's privileges.
CVSS v4.0 Base Metrics — Score 8.7 (HIGH)
Attack VectorNetwork
Attack ComplexityLow
Attack RequirementsNone
Privileges RequiredNone
User InteractionPassive
Confidentiality (Vulnerable System)High
Integrity (Vulnerable System)High
Availability (Vulnerable System)High
Confidentiality (Subsequent System)None
Integrity (Subsequent System)None
Availability (Subsequent System)None
CVSS v3.1 Base Metrics — Score 8.8 (HIGH)
Attack VectorNetwork
Attack ComplexityLow
Privileges RequiredNone
User InteractionRequired
ScopeUnchanged
ConfidentialityHigh
IntegrityHigh
AvailabilityHigh
Affected & Patched Versions
- huggingface datasets <= 5.1.0
- huggingface datasets 5.1.0