CVE-2025-5197 Regular Expression Denial of Service (ReDoS) in huggingface/transformers

CVE Details

Basic Information

Title CVE-2025-5197 Regular Expression Denial of Service (ReDoS) in huggingface/transformers
Type cve
Published 2025-08-06T11:53:37
Last Seen 2025-08-06T12:01:21
Modified 2025-08-06T11:53:37

CVSS Information

Base Score 5.3 (MEDIUM)
Attack Vector CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L
Confidentiality Impact NONE
Integrity Impact NONE
Availability Impact LOW

AI Analysis

AI Description A Regular Expression Denial of Service (ReDoS) vulnerability in the Hugging Face Transformers library’s `convert_tf_weight_name_to_pt_weight_name()` function can cause excessive CPU consumption due to catastrophic backtracking. This affects versions up to 4.51.3 and is fixed in version 4.53.0, potentially disrupting model conversion processes.
AI Severity Medium
AI Vendor Hugging Face
AI Product Hugging Face Transformers
AI Version up to 4.51.3

Additional Information

CVE List CVE-2025-5197
CWE List CWE-1333
Bulletin Family cve

Description

A Regular Expression Denial of Service (ReDoS) vulnerability exists in the Hugging Face Transformers library, specifically in the `convert_tf_weight_name_to_pt_weight_name()` function. This 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. The vulnerability affects versions up to 4.51.3 and is fixed in version 4.53.0. This issue can lead to service disruption, resource exhaustion, and potential API service vulnerabilities, impacting model conversion processes between TensorFlow and PyTorch formats.

💭 Join the Security Discussion

🔒 Your email address will not be published. Required fields are marked *

⚠️ Please be respectful and constructive in your comments. Security discussions should remain professional.