Sensitive Data Detection

Sensitive data detection identifies and protects personal information in text.

Overview

The sensitive data detection system identifies various types of personal information:

• Email addresses
• Phone numbers
• Credit card numbers
• Social security numbers
• IP addresses
• And more...

Configuration

Enable sensitive data detection:

guardrails:
  check_sensitive_data: true
  block_sensitive_data: true

Parameters

• check_sensitive_data: Enable/disable sensitive data detection
• block_sensitive_data: Whether to block content containing sensitive data

Usage

Automatic Detection

Sensitive data detection is automatic when enabled:

from elsai_guardrails.guardrails import LLMRails, RailsConfig

yaml_content = """
llm:
  engine: "openai"
  model: "gpt-4o-mini"
  api_key: "sk-..."

guardrails:
  check_sensitive_data: true
  block_sensitive_data: true
"""

config = RailsConfig.from_content(yaml_content=yaml_content)
rails = LLMRails(config=config)

result = rails.generate(
    messages=[{"role": "user", "content": "My email is user@example.com"}],
    return_details=True
)

if result.get('input_check'):
    sensitive = result['input_check'].sensitive_data
    print(f"Detected: {sensitive.get('predicted_labels', [])}")

Standalone Check

from elsai_guardrails.guardrails import GuardrailSystem, GuardrailConfig

config = GuardrailConfig(
    check_sensitive_data=True,
    block_sensitive_data=True
)
guardrail = GuardrailSystem(config=config)

result = guardrail.check_text("My email is user@example.com")
print(f"Safe to return: {result.passed}")

Result Format

Sensitive data results are returned in the sensitive_data field:

{
    'predicted_labels': ['Email']  # List of detected types
}

If no sensitive data is detected:

{
    'predicted_labels': ['No sensitive data detected']
}

Detected Types

Common sensitive data types include:

• Email
• Phone
• CreditCard
• SSN
• IPAddress
• And more...

Examples

No Sensitive Data

result = guardrail.check_text("Hello, how are you?")
print(result.passed)  # True

Email Detected

result = guardrail.check_text("My email is user@example.com")
print(result.passed)  # False
print(result.message)  # "Sensitive data detected."

Multiple Types Detected

result = guardrail.check_text("Email: user@example.com, Phone: 555-1234")
print(result.passed)  # False

Use Cases

Strict Mode (Block All)

guardrails:
  check_sensitive_data: true
  block_sensitive_data: true

Permissive Mode (Detect but Don't Block)

guardrails:
  check_sensitive_data: true
  block_sensitive_data: false  # Detect but allow

Detection Only (No Blocking)

guardrails:
  check_sensitive_data: true
  block_sensitive_data: false

Best Practices

1. Enable blocking in production to protect user privacy
2. Monitor detected data to understand what users are sharing
3. Consider use case - some applications may need to allow certain types
4. Provide clear feedback when sensitive data is detected

Next Steps

• Toxicity Detection - Detect toxic content
• Content Classification - Detect content issues
• Configuration Guide - Configure guardrails