Chrome extension that automatically detects AI-generated images, beauty filters, and photo manipulation — all offline, all client-side.
- AI-Generated Image Detection — 6-dimension analysis (frequency spectrum, noise, color, texture, symmetry, edges)
- Error Level Analysis (ELA) — Colored heatmap revealing tampered or edited regions
- Face Detection — BlazeFace-powered face detection + skin smoothness analysis
- Beauty Filter Detection — Detects skin smoothing filters via texture variance
- Screenshot Detection — Automatically skips screenshots and non-photo images
- Auto-Detection — Scans all images on any webpage with progress indicator
- Stamp Results — Marks suspicious images directly on the page (AI / Beauty Filter / Edited)
- Face-Only Mode — Only analyze images containing faces (default ON)
- Right-Click Analysis — Deep analysis of any image via context menu
- Settings Toggle — Enable/disable auto-detection and face-only mode
All processing happens locally in your browser. No images are uploaded. No data is sent to any server. Zero network requests. Your images never leave your device.
- Download the latest release and unzip
- Open
chrome://extensions/in Chrome - Enable Developer mode (top right)
- Click Load unpacked and select the unzipped directory
Coming soon
Re-compresses the image at a known quality level and compares the difference. Edited regions show higher error levels, visualized as a colored heatmap.
Uses BlazeFace to detect faces, then analyzes skin texture variance in face regions (forehead, cheeks, chin). Unnaturally low variance indicates beauty filter usage.
Scores images across 6 dimensions:
- Frequency Spectrum — AI images have steeper high-frequency falloff
- Noise Uniformity — AI images have unnaturally uniform noise distribution
- Color Distribution — AI images have smoother color histograms
- Texture Regularity — AI images have more regular LBP texture patterns
- Facial Symmetry — AI-generated faces tend to be more symmetrical
- Edge Coherence — Detects AI-typical edge artifacts
Identifies screenshots via flat color block ratio, color diversity, and noise level. Screenshots are automatically skipped in auto-detection mode.
- Manifest V3 — Modern Chrome extension architecture
- TensorFlow.js — Client-side ML inference
- BlazeFace — Real-time face detection model
- Canvas API — Image pixel manipulation and analysis
- Pure JavaScript — No build step, no frameworks
Contributions are welcome!
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
MIT License © 2026