Verify audio and images from your own apps

Quick start

Verify your first file in three steps.

1. 1

   Create a key

   Open the dashboard, name your key and choose an expiry. Copy it right away, it is shown only once.

2. 2

   Send a file

   POST the file as multipart/form-data to the verify endpoint with your key in the Authorization header.

3. 3

   Read the verdict

   The JSON response tells you whether the file is AI-generated (is_ai_generated) and how confident the model is (overall_confidence).

curl -X POST https://aiverification.com/api/v1/verify \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "file=@recording.mp3"

Authentication

Authenticate every request with your API key as a bearer token:

• Keys always expire (between 7 days and 1 year). Create a replacement before the old key runs out.
• You can have up to 5 active keys. Revoke a key at any time from the dashboard, and it stops working immediately.
• A key is shown only once when you create it. If you lose it, create a new one.

Verify endpoint

Supported files

• Audio, up to 40MBaudio/wavaudio/mp3audio/oggaudio/mpegaudio/m4aaudio/x-m4a
• Images, up to 20MBimage/jpegimage/pngimage/webp

Response

Every response is JSON with a success flag. On success the result is in data. A finished analysis returns HTTP 200.

chunk_details for audio

Audio is analyzed in chunks and listed in order, so you can see which part of a recording was flagged.

chunk_details for images

Example responses

{
  "success": true,
  "data": {
    "id": "3f2b8c1e-7a4d-4c1b-9e55-2d6f0a8b1c77",
    "status": "verified",
    "total_chunks": 2,
    "overall_score": 0.4837,
    "overall_confidence": 51.63,
    "is_ai_generated": true,
    "feedback": "NONE",
    "chunk_details": [
      {
        "score": 0.4837,
        "threshold": 0.5,
        "chunk_size": 30,
        "confidence": 51.63,
        "model_name": "AVI_AUDIO_V4",
        "prediction": "Fake",
        "time_seconds": 0.413
      },
      {
        "score": 0.4836,
        "threshold": 0.5,
        "chunk_size": 30,
        "confidence": 51.64,
        "model_name": "AVI_AUDIO_V4",
        "prediction": "Fake",
        "time_seconds": 0.336
      }
    ],
    "audio_length": 39.24,
    "created_at": "2026-10-22T02:22:45.20063+00:00",
    "verification_type": "audio",
    "flagged_by": "system",
    "original_file_name": "interview.mp3"
  }
}

How to read a result

• Use is_ai_generated for the verdict and overall_confidence for how sure the model is.
• Scores are raw model outputs. Do not compare them between audio and images.
• Results are probabilistic. For important decisions, combine them with other evidence.

Errors

Failed requests return a non-2xx status and a message you can show or log:

{
  "success": false,
  "error": {
    "message": "Invalid, expired or revoked API key"
  }
}

Limits