Vulnerability · Published Apr 7, 2026 · Assigned by @huntr_ai
CVE-2026-1839Arbitrary Code Execution via Unsafe torch.load() in Trainer Checkpoint Loading in huggingface/transformers
Medium 6.5
In plain English
A medium-severity unsafe deserialization in huggingface/transformers from huggingface. Proof-of-concept exploit code is public. The chance it is exploited in the next 30 days is estimated at 0.4% (higher than 33% of all known vulnerabilities). Fixed versions are available - upgrade the affected package.
Official 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.
Affected packages & fixed versions
| Ecosystem | Package | Fixed in | Advisory |
|---|---|---|---|
| PyPI | transformers | 5.0.0rc3 | GHSA-69w3-r845-3855 |
| PyPI | transformers | 5.0.0 | PYSEC-2026-2288 |
Affected products (as reported)
| Vendor | Product | Versions |
|---|---|---|
| huggingface | huggingface/transformers | unspecified – before v5.0.0rc3 |
Weakness: CWE-502 · Vector: CVSS:3.0/AV:L/AC:H/PR:N/UI:R/S:U/C:H/I:L/A:H