Skip to content
StartPrivate
Menu
Language: EN

Redact a screenshot

Find text spots email addresses, phone numbers, card and account numbers and QR codes; you box the names and avatars; export writes a new image without the hidden data of the original file.

Runs on your device. Files are not uploaded. How to check this yourself

or drop them here
Cover with

Face finder 0.9 MB · text finder 7.6 MB · each downloaded once, only when you press its button

    No file chosen yet.

      A screenshot often shows more than the message you meant to share: contact names, email addresses, phone numbers, profile pictures, links with private tokens. This tool reads the screenshot on your device to suggest boxes, covers them with solid blocks, and writes a new file, avoiding the 2023 bugs where cropped screenshots kept the hidden part inside the file.

      What screenshots give away

      Screenshots are the way people share proof: a funny message, a rude reply, a receipt, an error, a confirmation code. Because they capture the whole screen, they also capture everything around the point you are making:

      Cropping deals with the edges. Anything in the middle of the screen needs covering.

      Redact a screenshot step by step

      1. Take the screenshot as usual, then bring it here: file button, drag and drop, or on a phone, your photo library.
      2. Press Find text. The page reads the screenshot locally and boxes the email addresses, phone numbers, card and account numbers and QR codes it recognizes. The first time, it downloads its reader from this site and shows the size first.
      3. Box what the reader cannot know is private: names, usernames, profile pictures, addresses, order and account numbers, the address bar, other tabs.
      4. Make sure Cover with still says Solid block. Text under Pixelate or Blur can be pieced back together.
      5. Press Export, look over the copy at full size, and save or share it.

      How the text finding works

      Find text runs an OCR engine (Tesseract) inside your browser to turn the screenshot into words with positions, then applies simple rules:

      Screenshots are the best case for OCR: sharp, straight, high-contrast text. Small or light-gray text, text on photos, emoji-heavy names and unusual fonts can still be missed or misread. Treat the suggestions as a head start and look over the whole image before exporting.

      Why pixelated text in screenshots is a known weakness

      Screenshot text is rendered in a handful of standard system fonts at standard sizes, which is exactly what reconstruction tools need. Depix, published as a proof of concept in 2020, recovers text from screenshots that were pixelated with a simple box filter by generating candidate text in the same font, pixelating it, and matching the blocks. Its author notes the approach fails when the image is compressed afterwards, but you should not count on that. A solid block leaves nothing to match.

      The cropped-screenshot bugs of 2023

      Two widely used screenshot editors once leaked the parts you removed:

      Screenshots made and shared before those fixes can still carry the leftover data. Running them through this tool gives you a new file without that leftover: export paints only what is visible onto a new canvas and encodes a new image, so trailing data from the old file has nowhere to come from.

      What changes in the exported file

      Supported screenshots

      Where the screenshot goes

      It stays in your browser. Opening it reads the file into the tab; the OCR and barcode readers run in the tab; the finished copy waits in memory until you download it. Pressing Find text for the first time pulls the readers' program files from this site; those requests carry no part of your screenshot. Close the tab and the page's copies are gone.

      Technical details

      The OCR engine is tesseract.js with English language data, and square codes are found with ZXing compiled to WebAssembly; both are hosted on this site at fixed versions. The editor maps boxes from the preview back onto the original pixels, then encodes a new image. To see for yourself that no request carries your screenshot, follow the developer tools walkthrough.

      FAQ

      What should I hide in a chat screenshot?

      Names and profile pictures of everyone except you (unless they agreed), phone numbers, other chats' previews, and anything in a message that identifies a third person, like an address or workplace.

      Does Find text catch names?

      No. Names cannot be told apart from ordinary words by rules alone, so box them by hand. Find text covers email addresses, phone numbers, card numbers, IBANs, long digit runs and QR-type codes.

      Can someone recover text I covered?

      Not if you used Solid block: the exported file contains only the block. Pixelated or blurred text in screenshots is a known target for reconstruction tools.

      Is cropping enough?

      Cropping removes the edges, and in a buggy editor even that could leak. Here, export always writes a new file from what is visible, so leftovers from an earlier crop, or anything you covered, cannot hide inside it.

      Does the OCR upload my screenshot?

      No. Recognition happens inside your browser with an engine fetched from this site; the screenshot itself goes nowhere.

      Why redact the address bar?

      Links for password resets, shared files and some sign-ins carry secret tokens in the address. Anyone who can read the full link may be able to open it.

      Sources

      Checked 2026-10-11.

      Last updated: