Qencode
DEVELOPERS
All
All
API Reference
API Reference
Tutorials
Tutorials
Release Notes
Release Notes
Try QencodeSign In
HomeAPI ReferenceTutorialsRelease Notes
On this page
On this page

Detecting AI Generated Content

Introduction

The AI Detection feature analyzes video frames to estimate whether the video is AI-generated, returning a verdict and confidence score.

1
Start a Task with the "ai_detection" Output

AI Detection outputs can be created by adding an 'ai_detection' output format to your transcoding job. In order to do this, use the /v1/start_encode2 method to launch a transcoding job with the output parameter set to ai_detection.

Request Example

{
  "query": {
    "source": "https://nyc3.s3.qencode.com/qencode/bbb_30s.mp4",
    "format": [
      {
        "output": "ai_detection"
      }
    ]
  }
}

On the Transcode Media page choose AI Detection as an output format.

ai detection output

2
Access AI Detection

After the job is complete, you will be provided with a JSON file with the AI Detection verdict.

Output JSON structure:

{
  "total_frames": 5,                         
  "ai_frames": 3,                            
  "video_ai_ratio": 60.0,                    
  "verdict": "AI-GENERATED",                 
  "total_input_tokens": 6800,                
  "total_output_tokens": 475,                
  "errors": 0,                               
  "frames": [                                
    {
      "frame": "frame_0001.jpg",             
      "input_tokens": 1373,                  
      "output_tokens": 95,
      "is_ai": true,                        
      "reasoning": "The image displays...",  
      "video_ai_ratio": 60.0                 
    }
  ]
}
total_framesnumber of frames processed
ai_framesframes classified as AI
video_ai_ratioAI percentage (0-100)
verdict"AI-GENERATED" or "REAL"
total_input_tokenssum of input tokens across all frames
total_output_tokenssum of output tokens across all frames
errorsnumber of failed API calls
framefilename
input_tokenstokens consumed for this frame
output_tokenstokens consumed for this frame
is_aitrue/false/null (null on error)
reasoningmodel's analysis text (VERDICT line stripped)
video_ai_ratiosame as top-level, copied to each frame

3
Saving AI Detection output

You can specify the path for the output file in the destination.url parameter. It should include filename and extension.

Request Example:

{
  "query": {
    "format": [
      {
        "output": "ai_detection",
        "destination": {
          "url": "s3://us-west.s3.qencode.com/my-bucket/my-folder/ai_detection.json"
        }
      }
    ],
    "encoder_version": "2",
    "source": "https://my-bucket.media-storage.us-west.qencode.com/source_video.mp4"
  }
}

Set the path to the result file in the Output Path field. It should include filename and extension. The default value is ai_detection.json.

saving output

Need More Help?

For additional information, tutorials, or support, visit the Qencode Documentation pageLink or contact Qencode Support at support@qencode.com.