# Veil

> PII redaction middleware for LLM APIs

Veil is a drop-in privacy proxy that sits between your application and LLM providers (OpenAI, Anthropic, Google Gemini, Together, Groq, Mistral, DeepSeek, and 41 named providers). It automatically detects and redacts 79+ types of personal data before requests reach the LLM, then re-inserts original values into responses.

## Problem
Every LLM API call that includes user data (names, emails, SSNs, credit cards, API keys) sends that data to a third party. This violates GDPR, CCPA, HIPAA, and the EU AI Act (enforcement August 2026). Manual regex is fragile and misses evasion techniques.

## Solution
Change one URL in your OpenAI SDK configuration. Veil handles everything else:
- 79+ PII entity types across 18 countries
- API key and secret detection (AWS, GitHub, Stripe, GCP, JWTs)
- Crypto wallet detection (Ethereum, Bitcoin, Litecoin, Monero)
- Unicode evasion resistance (zero-width chars, Cyrillic homoglyphs)
- Real-time streaming with sliding-window restoration
- 23ms average overhead

## Quick Start
```python
from openai import OpenAI
client = OpenAI(
    api_key="your-openai-key",
    base_url="https://veil-api.com/v1",
    default_headers={
        "Authorization": "Bearer your-veil-key",
        "x-upstream-key": "your-openai-key",
    }
)
# That's it. PII is stripped automatically. Responses are restored.
```

## Pricing
- Free: 100 requests/month
- Starter: $49/month (10,000 requests)
- Growth: $149/month (100,000 requests)
- Enterprise: $499/month (1,000,000+ requests)

## Links
- API: https://veil-api.com
- GitHub: https://github.com/Atomics-hub/veil
- Docs: https://veil-api.com/docs