Bug 2559938 (CVE-2026-57845)
| Summary: | CVE-2026-57845 datasets: datasets: Arbitrary code execution via insecure pickle deserialization | ||
|---|---|---|---|
| Product: | [Other] Security Response | Reporter: | OSIDB Bzimport <bzimport> |
| Component: | vulnerability | Assignee: | Product Security DevOps Team <prodsec-dev> |
| Status: | NEW --- | QA Contact: | |
| Severity: | high | Docs Contact: | |
| Priority: | high | ||
| Version: | unspecified | CC: | alinfoot, bbrownin, dfreiber, drow, dtrifiro, ebourniv, hasun, jburrell, mhayden, mwoodson, nyancey, ptisnovs, sbunciak, sdoran, vkumar, weaton |
| Target Milestone: | --- | Keywords: | Security |
| Target Release: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | --- | |
| Doc Text: |
A flaw was found in datasets. This vulnerability allows an attacker to achieve arbitrary code execution by convincing a victim to load a maliciously crafted dataset file. Due to insecure deserialization—the process of restoring saved data back into program objects—untrusted data loaded through the pandas dataset builder is processed without validation. Consequently, opening a malicious file executes arbitrary Python code with the privileges of the running process.
|
Story Points: | --- |
| Clone Of: | Environment: | ||
| Last Closed: | Type: | --- | |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
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.