Redact a PDF

or drop one here
Choose a PDF first.

Draw boxes over private text and image areas, select text blocks, mark every match of a repeated term, or review detector suggestions, then download a rebuilt file with -redacted in its name. Marked content is removed from the page data and the pages are rebuilt from a fresh render, never just covered. The output must pass region, text, raw-data and privacy checks before the download appears.

The contract template going to a stranger

This page destroys content instead of hiding it. If the file leaves your hands and the names, numbers or faces inside must be unrecoverable, mark them here.

Say the sales team wants to reuse Acme-MSA-2025.pdf as a template for new prospects, but every page carries the old client's name, two IBANs and a scanned signature. You open it here on a Monday morning: Find suggestions flags the names and account numbers for review, Mark all matches catches the client's name on all eleven pages at once, and a drawn box takes the scanned signature image. Press Redact PDF and the template downloads with black boxes where the secrets were.

Four ways to mark: Draw a box over any area, Select text to grab whole text blocks, type up to 160 characters into Text to redact on every page and press Mark all matches, or press Find suggestions to run the 9.35 MB detector over selectable English text. Suggestions carry labels such as Person, IBAN or Date of birth, and every one waits for your review: nothing is marked until you accept it.

Two outputs: Keep unmarked text selectable leaves the rest searchable, while Flatten marked pages drops the text layer on those pages entirely for the strictest removal. Either way the pages are rebuilt from a fresh render, hidden document data is stripped, and the reopen check demands the same page count, exact surviving text and geometry, one black region per mark, zero raw fragments of removed words and zero remaining privacy findings. The result screen shows the same marked page before and after at the same zoom, so your eyes confirm what the checks proved.

Four steps from secrets to black boxes

  1. Choose a PDF

    Press Choose a PDF or drop the file onto the page. Locked files are refused; open them with Unlock PDF first.

  2. Mark private content

    Draw a box, Select text, Mark all matches for a repeated term, or Find suggestions and review each one. Select any mark to remove it; Undo, Redo, Clear page and Clear all sit below.

  3. Choose what remains selectable

    Keep unmarked text selectable for searchable documents, or Flatten marked pages when no text may survive on those pages.

  4. Redact and download

    Press Redact PDF. The rebuilt file is reopened and must pass every removal check before Download redacted PDF appears.

Redactions that leak, and how to close them

Review this list before you press Redact PDF:

The classic leak is a covered name that stays selectable: drawing a black box in an ordinary editor adds ink on top while the words survive underneath. This tool removes the words from the page data instead, which is why the reopen check hunts raw fragments. For an independent pass, run the download through Redaction check.

When destruction is the wrong answer

Redaction is one way and wide reaching: the output also loses form fields, comments, bookmarks and document metadata, because hiding places must go too. When the file must stay fillable or keep its history, send a copy with the sensitive pages removed instead, or share an excerpt.

Redaction cleans pages, not habits: a detector suggestion is a guess that can miss, especially outside English or in scanned images with no text layer. Read every page yourself before downloading. When the whole file is sensitive rather than a few regions, locking it with Password protect PDF beats blacking out half the pages.

Short answers about redacting PDFs

Can I redact a scanned signature?
Yes. Draw a box over the image area; the tool removes it from the rebuilt page. Text selection cannot grab image content, so boxes are the way for scans.
What does the detector miss?
Anything outside selectable English text, odd spellings, and content in images. It also caps at 250 pages and 2,000 suggestions. The review list exists because suggestions are a starting point, not a verdict.
Why did my form fields disappear?
Redaction strips form data, annotations, bookmarks and metadata along with the marked content, so nothing hides in the structures around the pages. Fill and flatten first when the answers must survive as print.
How do I prove the words are gone?
The reopen check requires zero raw fragments of removed words and zero privacy findings, and the result screen shows before and after at the same zoom. Run Redaction check on the download for a second opinion.
Can I redact only metadata, not pages?
This tool always rebuilds marked pages; for metadata alone, Remove PDF metadata is the lighter pass.

The rest of the kit, same local rules

Everything below runs in the browser with no account and reopens its own output before the download. Destroy here, verify with the checker, then scrub and send.