Detection API

Get started

1. Get AIPaperFix Pro. Pro & Pricing
2. Create an API key on your account page and keep it secret. Account
3. Send a POST request with your key in the Authorization header and the text as JSON.

Request

POST https://aipaperfix.com/api/v1/detect

curl https://aipaperfix.com/api/v1/detect \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text": "Paste the English or Chinese text you want to check here."}'

const response = await fetch("https://aipaperfix.com/api/v1/detect", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.AIPAPERFIX_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({ text }),
});
const result = await response.json();
console.log(result.ai_probability, result.label);

import os, requests

response = requests.post(
    "https://aipaperfix.com/api/v1/detect",
    headers={"Authorization": f"Bearer {os.environ['AIPAPERFIX_API_KEY']}"},
    json={"text": text},
    timeout=30,
)
result = response.json()
print(result["ai_probability"], result.get("label"))

Response

{
  "ai_probability": 0.82,
  "label": "ai",
  "sentences": [
    { "text": "First sentence of your text.", "ai_probability": 0.77 },
    { "text": "Second sentence.", "ai_probability": 0.88 }
  ],
  "engine": "model",
  "model_version": "aipf-2026.10-…"
}

ai_probability

Overall AI signal from 0 to 1.

label

"human", "mixed" or "ai" — the same three bands as the website (below 0.4, 0.4–0.7, above 0.7).

sentences

Each sentence with its own AI signal, in the original order.

engine

"model" normally. "rules" with "fallback": true means the detection model was briefly unavailable and a basic rule-based estimate was returned.

confidence

"low" when the text is too short for a reliable result.

Limits

• Up to 20,000 characters per request.
• Up to 60 requests a minute per key.
• English and Chinese text only.
• Up to 5 keys per account.

Errors

Errors return JSON like {"error": "…", "code": "…"} with these status codes:

About accuracy

Scores are estimates, not proof of who wrote a text. Rewritten or “humanized” AI text is often missed, and non-native and student writing can land in the mixed range. Don’t use a score as the only basis for academic, hiring or other important decisions. How it works