CVE-2024-5206

Sensitive data leakage in TfidfVectorizer due to improper token storage.

• CVSS 4.7
• CWE-922
• Information Leakage
• Local

A sensitive data leakage vulnerability was identified in scikit-learn's TfidfVectorizer, specifically in versions up to and including 1.4.1.post1, which was fixed in version 1.5.0. The vulnerability arises from the unexpected storage of all tokens present in the training data within the `stop_words_` attribute, rather than only storing the subset of tokens required for the TF-IDF technique to function. This behavior leads to the potential leakage of sensitive information, as the `stop_words_` attribute could contain tokens that were meant to be discarded and not stored, such as passwords or keys. The impact of this vulnerability varies based on the nature of the data being processed by the vectorizer.

CVE

CVE-2024-5206

CWE

CWE-922

CVSS base score

4.7

Published

2024-06-06

OWASP

A03 Sensitive Data Exposure

Orthogonal defect classification

Function

Code defect classification

Incorrect Functionality

Category

Information Leakage

Subcategory

Insecure Handling of Sensitive Data

Accessibility scope

Local

Impact

Information Disclosure

Affected component

scikit-learn

Fixed by upgrading

Yes

Solution

Upgrade to scikit-learn version 1.5.0 or later.

from sklearn.feature_extraction.text import TfidfVectorizer
import re

documents = [
    "Log entry: User 'admin' logged in with API key: sk-abcdefg1234567890",
    "Configuration file: password = mySecretPassword123",
    "Email: Please reset your password to: P@$$wOrd",
    "Normal text: This is a regular sentence."
]

vectorizer = TfidfVectorizer(stop_words='english')
X = vectorizer.fit_transform(documents)

stop_words = vectorizer.stop_words_

sensitive_patterns = [
    r"sk-[a-zA-Z0-9]+",
    r"[a-zA-Z0-9@#$%^&*()_+=-]+"
]

leaked_sensitive_data = []
for word in stop_words:
    for pattern in sensitive_patterns:
        if re.search(pattern, word):
            leaked_sensitive_data.append(word)

print("Stop words attribute:", stop_words)
print("\nPotentially leaked sensitive data:", leaked_sensitive_data)

print("\n--- Explanation ---")
print("This demonstrates CVE-2024-5206. The TfidfVectorizer, when processing text data, can mistakenly identify sensitive information (like API keys or passwords) as stop words.")
print("The 'stop_words_' attribute then exposes this sensitive data, creating a security vulnerability.")
print("An attacker could potentially extract this information and use it for malicious purposes.")

from sklearn.feature_extraction.text import TfidfVectorizer
import re

documents = [
    "Log entry: User 'admin' logged in with API key: sk-abcdefg1234567890",
    "Configuration file: password = mySecretPassword123",
    "Email: Please reset your password to: P@$$wOrd",
    "Normal text: This is a regular sentence."
]

sensitive_patterns = [
    r"sk-[a-zA-Z0-9]+",
    r"password\s*=\s*[^\s]+",
    r"(?i)password[^ ]*"
]

def sanitize(text):
    for pattern in sensitive_patterns:
        text = re.sub(pattern, "[REDACTED]", text)
    return text

sanitized_documents = [sanitize(doc) for doc in documents]

vectorizer = TfidfVectorizer(stop_words='english')
X = vectorizer.fit_transform(sanitized_documents)

print("TF-IDF matrix shape:", X.shape)
print("Sanitized input has been safely processed.")

print("\n--- Explanation ---")
print("This demonstrates CVE-2024-5206. The TfidfVectorizer, when processing text data, can mistakenly identify sensitive information (like API keys or passwords) as stop words.")
print("The 'stop_words_' attribute then exposes this sensitive data, creating a security vulnerability.")
print("An attacker could potentially extract this information and use it for malicious purposes.")

documents = [
    "User  password is: my_secure_password123",
    "API key for service: 12345-abcde-67890-fghij",
    "This is a normal document."
]

@misc{vaitp:cve20245206,
  title        = {{Sensitive data leakage in TfidfVectorizer due to improper token storage.}},
  author       = {Bogaerts, Fr\'ed\'eric and Ivaki, Naghmeh and Fonseca, Jos\'e},
  year         = {2024},
  note         = {VAITP Python Vulnerability Dataset, entry CVE-2024-5206},
  howpublished = {\url{https://netpack.pt/vaitp/vulnerability/CVE-2024-5206/}}
}