AI Image Detection — classify synthetic images

AI-Generated Image Detection API

What it does

AI Image Detection checks whether a publicly reachable JPG, PNG, or WebP image is likely AI-generated. Send an imageUrl and get back a verdict you can use to flag synthetic images in uploads, moderation flows, or content review pipelines.

The response includes a numeric ai_score from 0 to 1, a confidence level of low, medium, or high, and a verdict of likely_ai, likely_human, or uncertain. It also returns source_breakdown, which is currently an empty array and reserved for future per-generator scores.

Use AI Image Detection when you need a fast first pass on user-submitted imagery, marketplace listings, social posts, or verification flows where synthetic content should be reviewed differently from camera-captured images. The API is narrow by design: one image URL in, one classification result out.

Quickstart

Check whether an image URL looks AI-generated by sending the required imageUrl.

curl -X POST "http://localhost:8080/anyimagedetector/detect-ai-generated-image" \
  -H "apy-token: $APY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"imageUrl":"https://images.unsplash.com/photo-1503023345310-bd7c1de61c7d"}'

What you'll get back

Returns a JSON object with four top-level fields: verdict (string) indicating the detector's conclusion, ai_score (number) as the AI-likeness score from 0 to 1, confidence (string) for the confidence level, and source_breakdown (array) reserved for per-generator scores and currently empty.

{
  "verdict": "likely_ai",
  "ai_score": 0.87,
  "confidence": "high",
  "source_breakdown": []
}

About this endpoint

What it does

Detects whether a publicly reachable image is likely AI-generated. You send an image URL in the request body, and the response returns a verdict, AI score, confidence level, and a source breakdown array.

Request Body

Parameter	Type	Mandatory	Description
imageUrl	String	Yes	Publicly reachable JPG, PNG, or WebP URL. Format: URI.

Response

Returns a JSON object with four top-level fields: verdict as a string enum, ai_score as a number, confidence as a string enum, and source_breakdown as an array. The response uses these fields to report the image classification result and supporting breakdown data.

Parameter	Type	Mandatory	Description
verdict	ENUM	Yes	One of: `likely_ai`, `likely_human`, `uncertain`.
ai_score	Number	Yes	A numeric score between 0 and 1.
confidence	ENUM	Yes	One of: `low`, `medium`, `high`.
source_breakdown	Object Array	Yes	Reserved for future per-generator scores; currently empty.

Body

Name

Type

Description

bodyREQUIRED

object

—

Quickstart

What you'll get back

About this endpoint

What it does

Request Body

Response

Body

Errors any endpoint can return