your photos never leave this browser tab — we have no servers to send them to
⬤ ZERO TRANSMISSION — Face detection runs entirely inside your browser using WebAssembly. Before uploading, open F12 → Network tab and watch. You will see the model files load once from a CDN. Your photo data: never sent. Not to us. Not to anyone.
Loading face detection model from CDN...
Redact as
⬜
Drop photos here, or click to select
JPG · PNG · WEBP · processed locally · never uploaded
How it works
01
You pick a photo
File is read by your browser. Never leaves your device.
02
Model detects faces
SSD MobileNet v1 (WebAssembly) finds faces entirely in-browser.
03
Canvas redraws
Your browser draws the redaction shapes over the original image.
04
You download
A blob URL is created locally. The JPEG never touches a server.
Don't trust us — verify it yourself
  1. Open DevTools before touching anything: F12 on Windows/Linux, ⌘⌥I on Mac. Go to the Network tab.
  2. Upload a photo. Watch the Network tab in real time. You will see requests for model weights (ssd_mobilenetv1_model-*). That's it. Zero image data transmitted.
  3. All source code is readable right now: Ctrl+U / ⌘U → View Source. No obfuscation. No minification on the logic. Read every line.
  4. For absolute certainty: File → Save Page As → Webpage, Complete, then disconnect Wi-Fi. Open the saved file. It still works. There is no call home because there is nothing to call.
⬤ OFFLINE USE Save this page once using File → Save Page As → Webpage, Complete. After that first save, it works with zero internet connection — open the saved .html file directly from your hard drive. No servers involved because there are no servers. The model weights save alongside the HTML file automatically.