PII Removal

The PII Removal operation detects and removes personally identifiable information (PII) from messages to help ensure privacy compliance and data protection. It supports multiple PII types, replacement strategies, and format preservation options for comprehensive data sanitization.

Overview

Personally Identifiable Information (PII) poses significant privacy and compliance risks in AI systems. The PII Removal operation provides comprehensive protection by:

• Detecting Multiple PII Types - Emails, phone numbers, SSNs, credit cards, addresses, and more
• Flexible Handling Strategies - Redact, replace with realistic fake data, or remove entirely
• Format Preservation - Maintain document structure and readability while protecting privacy
• Compliance Support - Meet GDPR, CCPA, HIPAA, and other privacy regulations
• Custom Pattern Support - Define organization-specific PII patterns

The operation uses advanced pattern matching and contextual analysis to accurately identify PII while minimizing false positives that could disrupt legitimate content.

Configuration

Basic Configuration

const config = {
  types: ['email', 'phone', 'ssn', 'credit_card'],  // PII types to detect
  replacementStrategy: 'redact',                    // How to handle detected PII
  preserveFormat: true,                            // Preserve original format
  customPatterns: []                               // User-defined patterns
}

Configuration Options

Option	Type	Default	Description
types	array	['email', 'phone', 'ssn']	PII types to detect
replacementStrategy	string	'redact'	How to handle detected PII
preserveFormat	boolean	true	Preserve original format when replacing
customPatterns	array	[]	Custom regex patterns for organization-specific PII
confidenceThreshold	number	0.8	Minimum confidence score for PII detection

Supported PII Types

Type	Description	Detection Patterns
email	Email addresses	RFC 5322 compliant patterns
phone	Phone numbers	US/International formats
ssn	Social Security Numbers	XXX-XX-XXXX, XXXXXXXXX
credit_card	Credit card numbers	Visa, MasterCard, Amex, Discover
ip_address	IP addresses	IPv4 and IPv6 formats
address	Physical addresses	Street addresses with context
name	Person names	First/last name combinations
date_of_birth	Birth dates	Multiple date formats
passport	Passport numbers	Various country formats
license_plate	License plates	US state formats
bank_account	Bank account numbers	8-17 digit sequences
routing_number	Bank routing numbers	9-digit ABA numbers

Replacement Strategies

Strategy	Behavior	Example Output	Use Case
redact	Replace with [REDACTED]	Contact me at [REDACTED]	Maximum privacy protection
replace	Use realistic fake data	Contact me at jane@example.com	Maintain readability
remove	Delete PII entirely	Contact me at	Minimal data retention
mask	Partially obscure data	Contact me at j***@example.com	Balance privacy and context

Examples

REST API

curl -X POST https://api.meta-prompt.com/v1/process \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "messages": [
      {
        "role": "user", 
        "content": "Please contact John Smith at john.smith@company.com or call 555-123-4567. His SSN is 123-45-6789."
      }
    ],
    "operations": [
      {
        "name": "piiRemoval",
        "config": {
          "types": ["email", "phone", "ssn", "name"],
          "replacementStrategy": "redact",
          "preserveFormat": true
        }
      }
    ]
  }'

Response:

{
  "messages": [
    {
      "role": "user",
      "content": "Please contact [REDACTED] at [REDACTED] or call [REDACTED]. His SSN is [REDACTED]."
    }
  ],
  "operations": [
    {
      "name": "piiRemoval",
      "result": {
        "detectedPii": [
          {"type": "name", "value": "John Smith", "position": [14, 24]},
          {"type": "email", "value": "john.smith@company.com", "position": [28, 51]},
          {"type": "phone", "value": "555-123-4567", "position": [60, 72]},
          {"type": "ssn", "value": "123-45-6789", "position": [85, 96]}
        ],
        "replacementCount": 4
      }
    }
  ]
}

JavaScript SDK

import { MetaPrompt } from '@meta-prompt/sdk-js';

const client = new MetaPrompt('YOUR_API_KEY');

// Basic PII removal
const result = await client.process(
  [
    {
      role: 'user', 
      content: 'My credit card is 4111-1111-1111-1111 and email is user@domain.com'
    }
  ],
  [
    {
      name: 'piiRemoval',
      config: {
        types: ['email', 'credit_card'],
        replacementStrategy: 'redact'
      }
    }
  ]
);

console.log(result.messages[0].content);
// Output: "My credit card is [REDACTED] and email is [REDACTED]"

// Advanced configuration with custom patterns
const advancedResult = await client.process(
  [
    {
      role: 'user',
      content: 'Employee ID: EMP-12345, call me at john@company.com'
    }
  ],
  [
    {
      name: 'piiRemoval',
      config: {
        types: ['email'],
        replacementStrategy: 'replace',
        customPatterns: [
          {
            name: 'employee_id',
            pattern: 'EMP-\\d{5}',
            replacement: 'EMP-XXXXX'
          }
        ]
      }
    }
  ]
);

console.log(advancedResult.messages[0].content);
// Output: "Employee ID: EMP-XXXXX, call me at jane@example.com"

Python

Coming Soon!

This feature is currently under development and will be available shortly.