qwip AI Detection BETA

Name: qwip AI Detection BETA
Rating: 5 (5 reviews)
Author: Nathan Madsen

THIS EXTENSION IS FOR BETA TESTING. Privacy-first AI-generated image detection in real-time.

As of June 2026, qwip AI Detection BETA has 46 users and a 5.00/5 rating from 5 reviews in the Productivity category.

Nathan Madsen Productivity

Chrome Web Store ↗.crx

Users+360.0%

Rating0%

5.00

5 reviews

Reviews0%

Version

0.0.0.2

Manifest V3

History

8 snapshots

Tracking since Apr 6, 2026.

View as table

Date	Users	Rating	Reviews	Version
Apr 6, 2026	10	5.00	5	0.0.0.2
Apr 18, 2026	12	5.00	5	0.0.0.2
Apr 23, 2026	15	5.00	5	0.0.0.2
Apr 30, 2026	14	5.00	5	0.0.0.2
May 12, 2026	12	5.00	5	0.0.0.2
May 17, 2026	10	5.00	5	0.0.0.2
May 30, 2026	13	5.00	5	0.0.0.2
Jun 6, 2026	32	5.00	5	0.0.0.2
Now	46	5.00	5	0.0.0.2

Permissions & access

Permissions: offscreenstorageunlimitedStorageactiveTabcontextMenusalarms
Host access: <all_urls>

Screenshots

About

qwip is a privacy-first browser extension that detects AI-generated images entirely on your device. Your images never leave your browser.

🔒 PRIVACY-FIRST ARCHITECTURE
- 100% local processing using ONNX Runtime Web
- Zero server uploads - your images never leave your device
- No tracking, no accounts, no data collection
- Open source and auditable

🤖 ADVANCED AI DETECTION
- Multiple state-of-the-art models (MobileViT, Swin Transformer)
- Detects images from Stable Diffusion, DALL-E, Midjourney, and more
- ~85% accuracy on standard benchmarks (CiFake, GenImage)
- WebGPU acceleration for instant results

⚡ SEAMLESS BROWSER INTEGRATION
- Automatic detection of images as you browse
- Right-click any image to check AI detection
- Visual badges show results inline
- Works completely offline after initial download

🌐 COMMUNITY-POWERED DATABASE
- Optionally contribute anonymized perceptual hashes
- Help build global AI detection database
- Benefit from crowd-verified results
- All contributions are privacy-preserving (no images stored)

BETA TESTING NOTICE:
This extension is currently in beta. Accuracy will improve over time through model updates and community contributions. Detection results should be used as one factor in evaluating image authenticity, not as definitive proof.

TECHNICAL DETAILS:
- Uses ONNX Runtime Web with WebGPU acceleration
- Falls back to WASM on unsupported devices
- Perceptual hashing (pHash, BLAKE3) for duplicate detection
- Models: MobileViT CiFake, MobileViT GenImage, Swin AI Detector
- Total size: ~45MB (models downloaded on first run)

OPEN SOURCE:
GitHub: https://github.com/qwip-ai/qwip
Documentation: https://docs.qwip.io