mypdf.space

Guide

Redact a PDF before you share it

There is a recurring category of news story in which a government department or a law firm publishes a redacted document, someone selects the text underneath the black bars, and the unredacted version is everywhere within the hour. It keeps happening because of one misunderstanding.

Drawing a black rectangle in a PDF annotator adds a black rectangle. The text stays exactly where it was, one layer below, fully selectable and fully searchable. Real redaction has to delete the content, not cover it.

Step by step

  1. 1Run Privacy scanner over the document first to see where emails, phone numbers, card-like and ID-like strings appear.
  2. 2Open Auto-redact PII to mark the common patterns automatically, or Redact to draw over regions yourself.
  3. 3Work through every page, including headers, footers and anything in the margins.
  4. 4Apply the redaction and download the result — covered regions are flattened into the page, not layered over it.
  5. 5Open the downloaded file, try to select text through each black area, and search the document for a name or number you removed.
  6. 6Strip metadata with Remove metadata, since author names and file paths sit outside the page content entirely.

Where sensitive data hides

The visible page is the easy part. Document metadata carries the author name, the organisation, the software used and sometimes the full original file path on someone’s desktop — none of which is redacted by editing the page.

Then there are layers and annotations. Comments, sticky notes, form field values and previous edits can all persist in a file that looks clean. Flattening the document collapses these into the page so that what you see becomes all there is.

Attachments are the quiet one. A PDF can carry embedded files — a spreadsheet, the original source document — that never appear on any page. Check for them before sharing anything consequential.

Automatic scanning is a first pass, not a guarantee

A pattern scanner is very good at things with a shape: email addresses, phone numbers, card-like sequences, national ID formats. It will find instances you would have skimmed past on page forty of a long document, which is exactly what it is for.

It cannot recognise that a particular project codename is confidential, that one of the names listed is a protected witness, or that a date combined with a location identifies someone. Context-dependent sensitivity is a human judgement, so treat the scan as a starting checklist and then read the document yourself.

Verify before you send

Verification takes a minute and is the step people skip. Open the file you are actually about to send — not the editing session — and try to select text inside each redacted area. Nothing should highlight. Then search the document for a word, name or number you removed. It should return no results.

If either check fails, the content is still in the file. Better to discover that now than in a reply-all.

Tools used here

Frequently asked questions

  • Why is a black rectangle not enough?
    Because a PDF stores drawing instructions in layers. A rectangle is a new instruction on top; the text instruction underneath is untouched. Anyone can select it, copy it, or extract it with a script. Proper redaction removes the underlying content.
  • Is the redaction here permanent?
    Yes. When you apply it, the covered regions are flattened into the page so the original content is not present in the exported file. Verify on the downloaded copy by trying to select and search inside the redacted areas.
  • Can I redact a scanned document?
    You can draw redactions manually over any page, scanned or not. Automatic PII detection needs a text layer, so run OCR on a scan first if you want the scanner to find patterns for you.
  • What about the file name and the metadata?
    Neither is part of the page content, and neither is touched by redaction. Rename the file before sending it and use Remove metadata to clear the author, producer and path fields.

Other guides