AyeCalc
All tools
Image tools

AI Image Scanner: Check If an Image Is AI-Generated

Check whether an image contains visual patterns associated with AI generation. The scanner runs a quantized image-classification model in your browser, shows separate AI and real-image likelihood scores, and returns an inconclusive result when the signal is too close to call.

Image stays localCautious three-way verdictNo account
Browser-based visual classifier

Scan an image for AI patterns

Image stays local
Drop an image here

JPEG, PNG, or WebP · up to 25 MB and 25 megapixels

Private image processingThe selected image is analyzed inside this browser and is not uploaded for inference.

First-use downloadThe first scan downloads a quantized model of about 11 MB, plus browser runtime files.

Not definitive proofCompression, editing, screenshots, and unfamiliar generators can produce incorrect results.

Core formulaverdict = threshold(model AI-class score)

Scores of 70% or more are labeled likely AI-generated, 30% or less likely real, and the middle range inconclusive.

01
Method

How the ai image scanner works

After you start a scan, the browser downloads and caches a quantized Vision Transformer model, decodes the selected image locally, resizes and normalizes it for the model, and compares its output scores for the REAL and FAKE classes. A dedicated Web Worker runs the inference away from the main interface. The result is a classifier estimate based on learned pixel patterns, not verified authorship or provenance.

Formulaverdict = threshold(model AI-class score)

Scores of 70% or more are labeled likely AI-generated, 30% or less likely real, and the middle range inconclusive.

02
Worked example

Scan a compressed image downloaded from social media

Choose the image and start the scan. If the AI score is 58% and the real-image score is 42%, the tool reports Inconclusive rather than turning a narrow model difference into a confident claim. Compression and resizing may have removed signals the classifier learned during training.

03
Output

Using the processed image

Keep the exported dimensions, format, transparency, and compression level appropriate for where the image will be used.

Conservative thresholds
const verdict = aiScore >= 0.70
  ? 'likely-ai'
  : aiScore <= 0.30
    ? 'likely-real'
    : 'inconclusive';
Feature disclosure
<p>AI detection is an estimate, not proof of how an image was created.</p>
04
Practical guidance

Using the result accurately

Use the original, highest-quality image when possible. Treat the scanner as one signal alongside Content Credentials, source history, reverse-image search, contextual verification, and human review. Never accuse a creator, reject evidence, or make a consequential moderation decision from this score alone.

Important limitation

AI-image detectors can misclassify real photos, human-made digital art, edited images, screenshots, compressed files, and outputs from generators not represented in training. Metadata removal or the absence of Content Credentials does not prove an image is real or synthetic. The displayed scores are model outputs, not calibrated probabilities of authorship.

References: Hugging Face ONNX Community: distilled AI image detection model.

Common questions

AI Image Scanner FAQ

Answers about the method, assumptions, and practical use.

Can an AI image detector be 100% accurate?

No. Detection models can produce false positives and false negatives, especially for edited, compressed, resized, or unfamiliar images. This scanner deliberately reports an inconclusive middle range.

Does AyeCalc upload the image I scan?

No. The selected image is passed to a worker inside your browser for inference. The browser separately downloads model and runtime files on first use, but the selected image is not intentionally included in those requests.

Why does the first scan take longer?

The first scan downloads a quantized model of about 11 MB plus browser runtime files. The browser can cache those files, so later scans may start faster.

What does Inconclusive mean?

It means the model's AI score is between 30% and 70%, where the signal is not strong enough for this tool to label the image likely AI-generated or likely real.

Can the scanner identify which AI generator made an image?

No. This version estimates only whether the image resembles the model's real or AI-generated training classes. It does not reliably identify Midjourney, DALL-E, Stable Diffusion, or another specific source.