5.3

CVSS Score

3.0

-

CVSS Score

Basic Information

Concerned about an active attack path?

Talk to our security experts and see Miggo in action.

Miggo Vulnerability Database

→

CVE-2025-6921

CVE-2025-6921: Hugging Face Transformers vulnerable to Regular Expression Denial of Service (ReDoS) in the AdamWeightDecay optimizer

The huggingface/transformers library, versions prior to 4.53.0, is vulnerable to Regular Expression Denial of Service (ReDoS) in the AdamWeightDecay optimizer. The vulnerability arises from the _do_use_weight_decay method, which processes user-controlled regular expressions in the include_in_weight_decay and exclude_from_weight_decay lists. Malicious regular expressions can cause catastrophic backtracking during the re.search call, leading to 100% CPU utilization and a denial of service. This issue can be exploited by attackers who can control the patterns in these lists, potentially causing the machine learning task to hang and rendering services unresponsive.

(GitHub Advisory)

Miggo Vulnerability Database

→

CVE-2025-6921

CVE-2025-6921:

5.3

CVSS Score

3.0

-

CVSS Score

Basic Information

Is this CVE running in your environment?

Easily map the attack path and prioritize which CVEs are a threat to your organization

Validate Exposure

Technical Details

Package Name	Ecosystem	Vulnerable Versions	First Patched Version
transformers	pip	< 4.53.0	4.53.0

Technical Details

Vulnerability Intelligence
Miggo AI

Root Cause Analysis

The analysis of the provided security patches, specifically commit d37f7517972f67e3f2194c000ed0f87f064e5099, reveals two instances of Regular Expression Denial of Service (ReDoS) vulnerabilities within the Hugging Face Transformers library. The primary vulnerability, as described in the CVE, is located in the _do_use_weight_decay method of the AdamWeightDecay optimizer. This function insecurely used re.search with user-provided patterns, making it susceptible to malicious regular expressions that can cause the application to hang. The patch replaces this unsafe call with a simple substring check. A similar vulnerability was identified and fixed in the remove_language_code method of the MarianTokenizer class, where an inefficient regular expression was replaced with basic string operations. Both functions would appear in a runtime profile during an exploit, as they are directly involved in processing the malicious input that triggers the denial of service.

Vulnerable functions

Only Mi**o us*rs **n s** t*is s**tion

Vulnerability Intelligence
Miggo AI

Unlock WAF rules for this CVE

Generate vendor-ready rules for the observed attack patterns, plus reasoning and safe deployment guidance

Get WAF rules

WAF Protection Rules

WAF Rule

W** rul*s *v*il**l* *or Mi**o *ustom*rs only.W** rul*s *v*il**l* *or Mi**o *ustom*rs only.W** rul*s *v*il**l* *or Mi**o *ustom*rs only.W** rul*s *v*il**l* *or Mi**o *ustom*rs only.W** rul*s *v*il**l* *or Mi**o *ustom*rs only.W** rul*s *v*il**l* *or Mi**o *ustom*rs only.W** rul*s *v*il**l* *or Mi**o *ustom*rs only.W** rul*s *v*il**l* *or Mi**o *ustom*rs only.W** rul*s *v*il**l* *or Mi**o *ustom*rs only.W** rul*s *v*il**l* *or Mi**o *ustom*rs only.

Reasoning

*v*il**l* *or Mi**o *ustom*rs only.*v*il**l* *or Mi**o *ustom*rs only.*v*il**l* *or Mi**o *ustom*rs only.*v*il**l* *or Mi**o *ustom*rs only.*v*il**l* *or Mi**o *ustom*rs only.*v*il**l* *or Mi**o *ustom*rs only.*v*il**l* *or Mi**o *ustom*rs only.*v*il**l* *or Mi**o *ustom*rs only.*v*il**l* *or Mi**o *ustom*rs only.*v*il**l* *or Mi**o *ustom*rs only.