Redact a PDF

Drag a black box over the data you want to hide and download a PDF where that text no longer exists: the page is rebuilt as an image, so there is nothing left under the black to select or copy. Your file never leaves your device.

📄 Drop your PDF here, or click to choose it

It stays on your device: no upload, no server.

🔒 The PDF is opened and rewritten inside your browser: it is never sent anywhere.

Why the black box you draw in Word or Preview hides nothing

In a PDF, text and graphics are separate objects stacked on top of each other. When you add a black rectangle in Word, with the shape tool of macOS Preview or in any reader, you are only putting an object on top of the letters: the text is still in the file, exactly where it was. Anyone can select and copy it, open the PDF in another program, extract the text, or simply delete that rectangle and read the account number you thought was covered. It is one of the classic privacy accidents, and it has happened to courts and government offices too. To be fair: Preview also has a dedicated redaction feature (Redact, introduced with macOS Big Sur) that really does remove the content, that is a different thing from a hand-drawn rectangle, and it is not what this page is warning you about.

What this tool does instead: every page is rebuilt as an image

Here the box is never laid on top of the PDF. When you press the button, each page is redrawn onto a canvas at the resolution you picked, the boxes are painted straight onto the pixels, and the result is saved as a JPEG image; all the images are then packed into a brand new PDF built from scratch. The file you download contains no text objects at all: under the black there is nothing to select, copy or extract, because those pixels are simply black. The metadata of the original PDF (author, producing software, revision history) is left behind as well: the new file only carries the name of the library that wrote it and the date you generated it.

How to use it

Load the PDF, drag a box over each piece of data you want gone, and undo the last one or clear the page if you misdraw. Boxes stay attached to their own page, so you can move back and forth and pick up where you left off. Choose the quality (150 dpi is fine almost always; pick 200 dpi if the print is small or you need scan-like quality), press the button and watch the progress bar, a twenty page document takes a few seconds, because every page really is redrawn. One tip: leave a margin around the data, since accents and descenders stick out further than you expect.

The limits, stated plainly

The PDF you get is made of images: it is heavier than the original, it is no longer searchable or selectable (not just in the covered areas, on the whole page), you cannot copy and paste from it, and a screen reader can no longer read it. The new file contains only the pages as pictures: clickable links, bookmarks, comments and notes (of which at most the drawing survives, never the text) and any attachments are gone too. If the document is digitally signed the signature is broken, and if it has fillable forms the fields become a picture (the value typed in them stays visible, but as an image): the tool warns you before you download, checking both the document metadata and the usual traces (/ByteRange, /AcroForm) in the raw bytes, an indicative check, not a validation of the signature itself. PDFs that ask for a password to open cannot be opened at all: remove the password first (files protected only against printing or copying do open, and the redacted copy will not inherit those restrictions). Finally, on a phone with little memory a very long document at high resolution may fail: pick 100 dpi in that case.

Always check the result before you send it

The check that matters is done with your eyes: open the downloaded PDF and go through every page, making sure the black covers each item completely and that the item does not show up somewhere else, in a header, in a summary, on a page you never looked at. Beware of one misunderstanding: searching with Ctrl+F (or cmd+F) will never find anything, and not because you covered it well, but because the whole page is now an image, that test only proves the text can no longer be extracted, not that you covered the right thing. Finally, remember that hiding data here does not erase it from your original: keep the original safe and send only the redacted copy.