CVE-2025-46567 json
LLaMA-Factory Allows Arbitrary Code Execution via Unsafe Deserialization in Ilamafy_baichuan2.py
published 2025-04-23 18 mo ago modified 2026-09-10 1 mo ago also GHSA-f2f7-gj54-6vpv , PYSEC-2026-1574 , EUVD-2025-12156
- cvss
- 6.1
- version 3.1, CNA
- cisa kev
- Not listed
- not in CISA's catalog
- epss
- 0.3%
- EPSS 0.3% chance of exploitation in 30 days, percentile 21%
- public exploits
- 0
- none known
Summary
LLama Factory enables fine-tuning of large language models. Prior to version 1.0.0, a critical vulnerability exists in the `llamafy_baichuan2.py` script of the LLaMA-Factory project. The script performs insecure deserialization using `torch.load()` on user-supplied `.bin` files from an input directory. An attacker can exploit this behavior by crafting a malicious `.bin` file that executes arbitrary commands during deserialization. This issue has been patched in version 1.0.0.
Patch / fix
CISA SSVC
- Exploitation
- Proof of concept is public
- Automatable
- no
- Technical impact
- partial
- Assessed
- 18 mo ago
Exploit prediction
0.29% chance of exploitation in the next 30 days, higher than 21% of all scored CVEs (FIRST EPSS, 2026-11-08).
Exploits and checks
- github.com https://github.com/hiyouga/LLaMA-Factory/security/advisories/GHSA-f2f7-gj54-6vpv referenced PoC, tagged "exploit" by a source, unverified
CVSS scores
| assigned by | version | score | vector |
|---|---|---|---|
| CNA | 3.1 | 6.1 | CVSS:3.1/AV:L/AC:L/PR:L/UI:R/S:U/C:H/I:L/A:L |
| NVD | 3.1 | 7.8 | CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H |
| GitHub | 3.1 | 6.1 | CVSS:3.1/AV:L/AC:L/PR:L/UI:R/S:U/C:H/I:L/A:L |
| OSV | 3.1 | 6.1 | CVSS:3.1/AV:L/AC:L/PR:L/UI:R/S:U/C:H/I:L/A:L |
Affected packages
| ecosystem | package | affected | fixed in |
|---|---|---|---|
| pypi | llamafactory | <0.9.3 |
0.9.3 |
Affected products
- hiyouga LLaMA-Factory
Classification
- Weakness
-
- CWE-502 Deserialization of Untrusted Data
References
-
github.com
/hiyouga/LLaMA-Factory/commit/2989d39239d2f46e584c1e1180ba46b9768afb2a
web, patch
-
github.com
/hiyouga/LLaMA-Factory/security/advisories/GHSA-f2f7-gj54-6vpv
exploit, web, vendor-advisory
-
nvd.nist.gov
/vuln/detail/CVE-2025-46567
advisory
-
github.com
/hiyouga/LLaMA-Factory
package
-
github.com
/hiyouga/LLaMA-Factory/blob/main/scripts/convert_ckpt/llamafy_baichuan2.py#L35
web
-
pypi.org
/project/llamafactory
package
-
github.com
/advisories/GHSA-f2f7-gj54-6vpv
advisory
Sources
- CVE List CVE-2025-46567
- GitHub Advisory GHSA-f2f7-gj54-6vpv
- NVD CVE-2025-46567
- OSV PYSEC-2026-1574
Fencer Vulns keeps one entry per vulnerability: the CVE id wins, then GHSA, and records that alias each other are merged, except that two CVEs always keep separate entries. The same data is available as JSON.