PDF Redact covers names, phone numbers, ID numbers, amounts and other sensitive content for good — in files you have the right to handle. Mark keywords in the text layer in bulk, or draw boxes by hand in the page preview. When you run it, every marked page is rasterised whole, the black blocks are baked straight into the pixels, and the output file drops document metadata such as title, author, subject and keywords.

This is not a black rectangle that someone can delete later. A marked page becomes an image page and the original text layer no longer exists, so the covered area cannot be recovered by copying text or peeling off an overlay. Pages you did not mark are copied through unchanged.

1. 01Choose a file you can readselect the PDF on the left; if it has an open password, unlock it or enter the password so the content becomes readable.
2. 02Mark the sensitive areaswhen a text layer exists, type terms into “Keyword search” and decide whether matching is case-sensitive; for a scan with no searchable text, click “Open page marking preview” and draw the boxes by hand.
3. 03Check page by pagesearch hits are added to the marks automatically and the view jumps to the first one; in the preview, inspect the boundaries, add anything missed, delete wrong marks, and confirm the number of marks and the pages they cover.
4. 04Choose the output resolution and runpick 150, 200 or 300 DPI and click “Redact and download”; marked pages are rasterised with the blocks written in, and the file is saved as <original-name>_redacted.pdf.

Keyword search

searches only the text layer that PDF.js can read; scanned images, outlined text or text with a broken encoding may not be found, so fall back to manual boxes.

The cost of the output

a marked page is rasterised whole, so its text can no longer be selected, copied or searched; the higher 150 / 200 / 300 DPI usually means a sharper but larger file.

Page order

marks follow the original file's page order, independent of any rearrangement in the workspace.

Privacy review

the tool strips the common document metadata from the output, but you should still check every page for marks you missed, QR codes, images or other identifying content.

What it does not do

it does not restore redacted content, does not decide by itself what counts as sensitive, and does not replace your organisation's review process; only work on files you have the right to edit and disclose.

Reading the PDF, locating keywords, rendering pages and burning in the redactions all happen locally in the browser, and the file is never uploaded. The output also has common document metadata removed, but review it by hand before publishing to be sure nothing sensitive was missed.