Jev in production › Moderation and content filtering

Slop Scope

Browser-run AI image detector sends measured image facts to a Jev model, the sole classifier labeling images likely_ai, likely_real or uncertain.

Open on GitHub ↗

The source publishes no measured number.
Use
Moderation and content filtering
Industry
Media and social
Form
Open-source tool
Stage
Announced
Listed
2026-10-04
Found via
github
Repository
GPTchatly/slop-scope
Stars
3
Forks
0
Last push
2026-10-03
Language
JavaScript
License
MIT

The README opens with

Slop Scope asks whether an image is AI-generated. Your browser measures the image and a JEV model judges the measured facts. This edition runs on your own computer with your own OpenRouter key. It needs no browser check and no external database, and it enforces no usage limits.

- The browser does everything that touches the image. It parses and decodes the file. It measures 16 windows and the lens-fringe field. It runs our whole-image detector, pixel-detector-5-f1: a CLIP ViT-B/16 vision transformer, fine-tuned with JEV-weighted training. The vendored onnxruntime-web WebAssembly runtime runs it. Then the browser derives a fixed fact dossier. - Only the fixed dossier vocabulary leaves the browser. It goes to the server on this computer as an...

Badge

For the project's own README, linking back here:

Listed in Jev in production

Also used for moderation and content filtering