CVE-2026-1839
A vulnerability in the HuggingFace Transformers library, specifically in the `Trainer` class, allows for arbitrary code execution.
Does this matter?
High impact if exploited, but EPSS currently rates exploitation as unlikely (0.35%). Schedule it in the normal patch cycle and watch for a rise in EPSS or a public exploit.
Description
A vulnerability in the HuggingFace Transformers library, specifically in the `Trainer` class, allows for arbitrary code execution. The `_load_rng_state()` method in `src/transformers/trainer.py` at line 3059 calls `torch.load()` without the `weights_only=True` parameter. This issue affects all versions of the library supporting `torch>=2.2` when used with PyTorch versions below 2.6, as the `safe_globals()` context manager provides no protection in these versions. An attacker can exploit this vulnerability by supplying a malicious checkpoint file, such as `rng_state.pth`, which can execute arbitrary code when loaded. The issue is resolved in version v5.0.0rc3.
- CVSS 3.1
- 7.8 HIGHCVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
- EPSS
- 0.35% probability · 28th percentile
- CISA KEV
- Not listed
- Weakness
- CWE-502
- Affected
- huggingface/transformers
- Source
- security@huntr.dev
References
- https://github.com/huggingface/transformers/commit/03c8082ba4594c9b8d6fe190ca9bed0e5f8ca396Patch
- https://huntr.com/bounties/3c77bb97-e493-493d-9a88-c57f5c536485Exploit, Third Party Advisory
- https://huntr.com/bounties/3c77bb97-e493-493d-9a88-c57f5c536485Exploit, Third Party Advisory
Source: NVD record, EPSS from FIRST.org, KEV from CISA. Refreshed daily.