Paste (Ctrl+V), drop or click to load an imageBlack out private details in a screenshot. Finding them automatically downloads 3.5 MB the first time.

Redact a screenshot online

Black out names, emails, account numbers, API keys or anything else private in a screenshot. The covered areas are flattened into the saved file, so the hidden content is not in the image you share. For text, a solid box is safer than blurring part of the image or pixelating the area, since both can be read back.

How to redact a screenshot#

  1. Paste a screenshot with Ctrl+V, drop a file on the page or press Load.
  2. The Shape tool is ready to draw solid rectangles. Drag a box over each detail to hide; pick black (or any colour) from the colour swatch.
  3. Repeat for every sensitive area. Use the undo button, Ctrl+Z or History to undo a box.
  4. Copy with Ctrl+C, or save as PNG, JPEG or WebP.

The Enhance tool can leave traces of what boxes and blurs hide. Each pixel it makes depends on the pixels around it, those under a box too. To remove the noise from a photo you're redacting, or to upscale the image, save the redacted image, open the saved file and use the tool there.

Find sensitive info automatically#

Load a screenshot and press the Find button that appears once it's open, or the Find sensitive info button on the bar. Text recognition reads the image on your device, as it does to copy text from an image, and outlines whatever looks like an email address, a phone number, a card number, an IBAN, an IP address, an API key or token, or a link with a password or token in it.

What automatic detection misses#

It doesn't look for names, postal addresses, signatures or ID photos, and faces have a finder of their own: blur or pixelate faces. It finds a password inside a link, in a line of code or settings such as DB_PASSWORD=…, or when it's long and random. A password in a form field, or a plain word after a label such as “Wi-Fi password: bluefish”, is missed. It reads clear printed text best. Handwriting, very small or faint text and text on a busy background can slip through. Cover those by hand.

Tap an outline to leave it out, or a kind of value in the bar to leave out all of that kind, then press the button with the count, such as Redact 7, to cover the rest in one step. Check the result before sharing. Detection can miss things.

A screenshot of account settings with black boxes over the name, the email address, the phone number, the IBAN, the card number, both API keys, the IP address and the avatar's initials.
Seven values found and covered with the Redact 7 button, and the name and its initials with boxes drawn by hand.

Download size#

With no model files stored, text recognition needs 3.5 MB for its model and runtime. The page asks before downloading missing files.