Overview
A critical vulnerability, identified as CVE-2025-2099, has been discovered in the `transformers.testing_utils` module of huggingface/transformers, a popular machine learning library. This vulnerability, specifically within the `preprocess_string()` function, potentially exposes systems to Regular Expression Denial of Service (ReDoS) attacks. It is significant as it can lead to high system CPU usage, resulting in potential application downtime and posing a risk to system stability and data security.
Vulnerability Summary
CVE ID: CVE-2025-2099
Severity: High (7.5 CVSS Score)
Attack Vector: Network
Privileges Required: None
User Interaction: None
Impact: Potential system compromise or data leakage
Affected Products
Escape the Surveillance Era
Most apps won’t tell you the truth.
They’re part of the problem.
Phone numbers. Emails. Profiles. Logs.
It’s all fuel for surveillance.
Ameeba Chat gives you a way out.
- • No phone number
- • No email
- • No personal info
- • Anonymous aliases
- • End-to-end encrypted
Chat without a trace.
Product | Affected Versions
huggingface/transformers | v4.48.3
How the Exploit Works
The vulnerability exists in the `preprocess_string()` function of the `transformers.testing_utils` module in huggingface/transformers. The regular expression used for processing code blocks in docstrings has nested quantifiers. This causes exponential backtracking when processing input with a large number of newline characters. An attacker can exploit this by providing a specially crafted payload, causing high CPU usage and potential application downtime. This effectively allows for a Denial of Service (DoS) scenario.
Conceptual Example Code
The following pseudocode example demonstrates how an attacker might exploit this vulnerability:
import transformers.testing_utils as utils
malicious_payload = "\n" * 100000 # A long string of newline characters
utils.preprocess_string(malicious_payload)In this conceptual example, the `malicious_payload` string consists of a large number of newline characters. When passed to the `preprocess_string()` function, it triggers the vulnerability, leading to high CPU usage and potential denial of service.
