SB2026021756 - Inefficient regular expression complexity in Hugging Face Transformers library

Description

This security bulletin contains information about 1 security vulnerability.

1) Inefficient regular expression complexity (CVE-ID: CVE-2025-5197)

The vulnerability allows a remote attacker to perform a denial of service (DoS) attack.

The vulnerability exists due to `convert_tf_weight_name_to_pt_weight_name()` function responsible for converting TensorFlow weight names to PyTorch format, uses a regex pattern `/[^/]*___([^/]*)/` that can be exploited to cause excessive CPU consumption through crafted input strings due to catastrophic backtracking. A remote attacker can pass specially crafted data to the application and perform regular expression denial of service (ReDos) attack.

Remediation

Install update from vendor's website.

References

• https://github.com/huggingface/transformers/commit/944b56000be5e9b61af8301aa340838770ad8a0b
• https://huntr.com/bounties/3f8b3fd0-166b-46e7-b60f-60dd9d2678bf