Skip to content

Privacy

The image never leaves this tab.

Not after processing. Not for moderation. Not as an anonymous analytics event. Your browser reads the file, redacts it, and writes a new one.

Open the editorVerify it yourself

This page explains how the editor handles your image. For the formal policy — cookies, hosting logs, and your rights — read the Privacy Policy.

Privacy Policy
No image uploads
No account
No analytics
No tracking cookies

Data flow

The whole trip happens inside the page.

There is no processing queue, no temporary bucket, and no “we delete it after 24 hours” promise to trust. The file is decoded by your browser.

In this browser tab

Local only
  1. 01

    You choose a file

    A file picker, a paste, or a drop. The browser creates a local object URL — no request leaves the page.

  2. 02

    The browser decodes it

    The image is drawn to an in-memory canvas. This is the same canvas API every image editor uses.

  3. 03

    You draw the boxes

    Redactions are drawn over the local pixels. Undo history lives in JavaScript memory, not a database.

  4. 04

    Canvas exports a new image

    The download is a fresh file encoded from the canvas. It carries no EXIF, GPS, or device metadata.

No server in the middle.

There is no endpoint to receive an image, so there is nothing to retain, log, moderate, or sell. Closing the tab clears the working copy from memory.

Illustrative request

POST /upload
→ 404 Not Found

No such endpoint exists on imageredaction.com.

Network requests

The honest list.

Smart detection is optional, and it is the only feature that needs to fetch code after the page loads. Here is every request the editor can make.

This page and its font

When: When you open the site

HTML, CSS, JavaScript, and the self-hosted Inter font. No image.

Tesseract.js

When: First time you run Smart detect

The OCR library from jsDelivr. No image.

OCR core and English model

When: First Smart detect in a browser

WebAssembly and language data requested by Tesseract.js. No image.

Your screenshot

No uploads.

When: Never

Nothing. There is no upload endpoint to receive it.

What “on-device” means here

The OCR engine is downloaded code, and the recognition runs inside your browser. Your image is not sent with the request. The static host and the CDN may keep ordinary access logs for the requests that do happen, as any website or CDN does. Those logs contain the request URL, not your screenshot.

Verify it yourself

Do not take the page’s word for it.

The claim is testable in about two minutes. If you find a request that contains your image, that is a bug worth reporting.

Report a problem
  1. 1

    Open your browser’s developer tools and switch to the Network tab. Clear the list.

  2. 2

    Load an image and draw a few boxes. Watch the request list: no image request appears.

  3. 3

    Run Smart detect. You will see requests for the OCR library and model. The screenshot is not among them.

  4. 4

    Disconnect from the internet. Manual redaction, copy, and download still work because the pixels are already local.

  5. 5

    Open the downloaded file’s properties. There is no camera, GPS, or device metadata field to find.

Your device, your responsibility

Browser extensions, clipboard managers, and screen-recording tools can read what happens in a tab. If your threat model includes a compromised browser, use a clean profile or a separate device.

No account, no profile

There is no sign-in, no email capture, and no per-user storage. The editor has no idea who you are. That is also why there is no usage history to export.

Open to inspection

The editor is plain JavaScript and canvas. There is no hidden native layer. Use the network tab, read the request list, and test the offline behaviour.

Privacy questions

The short version, repeated.

The same answers as the FAQ, focused on where the image goes and what leaves the page.

Are my images really never uploaded?

Correct. Your browser reads the file, draws it to a local canvas, and keeps it in memory. There is no upload endpoint — you can watch the network tab while you work. Smart detection downloads an OCR engine and an English text model on first use, but the image itself stays in the tab.

Is it free? Do I need an account?

It is free, with no account and no watermark. Open the editor and use it. There is no paid export, no credit counter, and no reason to hand over an email address before you start.

How does smart detection work?

It runs Tesseract OCR in a Web Worker inside your browser, then applies pattern checks: emails, Luhn-valid card numbers, SSNs, Aadhaar numbers, PANs, GSTINs, IBANs, phone numbers, IP addresses, dates, address keywords, and likely names. Every finding is a suggestion you can include or exclude. OCR is best-effort, so review the boxes and use Verify.

Does it work offline?

The editor does. Once the page is open, you can keep redacting without a network connection. Smart detection needs the OCR engine and English model on first use; your browser may cache them, but do not rely on that for a first offline run. Manual redactions, copy, and export all work offline.

Is EXIF and GPS data removed?

Yes. Canvas export re-encodes the pixels, so camera metadata, GPS coordinates, device details, and the original file name do not travel with the new file. Screenshots often have no EXIF to begin with; re-encoding also normalises PNG, JPG, and WebP exports.

Can I make the exported file smaller?

Yes. The Size panel in the editor measures the real export before you download it: lower the quality for JPG and WebP, or cap the longest edge to resize the flattened image. You see the actual byte count and the percentage saved. PNG is lossless, so quality does not apply to it — use the size cap or choose WebP. Redactions stay baked in either way.

See for yourself

Bring an image. Watch the network tab.

The editor is the proof. Load a screenshot, redact it, export it, and inspect the file. Nothing needs to leave the tab.

Open the editorRead the guides

Free · No account · Nothing uploaded