Constitutional Verification API

Evaluate, govern, and certify any AI output against constitutional principles. Free. Open. Infrastructure.

What this does

The Article 11 API runs every input through 6 constitutional verification stages. It checks AI outputs against 5 governing principles, 52 injection patterns across 11 OWASP categories, and returns human-readable explanations for every flag. Not just "blocked" — why it was blocked and how to fix it.

This is what other AI safety systems don't do. They say "content filtered." We say which article you violated and which principle applies.

Base URL

https://api.article11.ai/api/v1

Endpoints

{ "text": "The AI output you want to verify" }

{ "prompt": "The user's input", "response": "The AI's response to verify" }

{ "prompt": "The user's input", "response": "The AI's response to certify" }

Try it now

Quick start

# Evaluate any text
curl -X POST https://api.article11.ai/api/v1/evaluate \
  -H "Content-Type: application/json" \
  -d '{"text":"Your AI output here"}'

# Verify a prompt+response pair
curl -X POST https://api.article11.ai/api/v1/govern \
  -H "Content-Type: application/json" \
  -d '{"prompt":"User question","response":"AI answer to verify"}'

# Get a signed certification on the IRONLEDGER
curl -X POST https://api.article11.ai/api/v1/certify \
  -H "Content-Type: application/json" \
  -d '{"prompt":"User question","response":"AI answer to certify"}'

The five principles

Pricing

Architecture

The Constitutional Intelligence Pipeline (CIP) runs 6 verification stages on every request. Safety lives in the infrastructure — outside the AI model, in the Cloudflare Worker. The model cannot override the Worker.

The Constitution is CC0. Read it, fork it: article11.ai/constitution