How to redact a PDF properly

The short answer

Use a tool that removes the text, not one that covers it. Open your PDF in pdfanonym's redaction tool, select what must go (or search for it), press Apply, download, then check the result by trying to copy and search the redacted words. If nothing comes back, the redaction worked.

The rest of this guide explains why that check matters and how to deal with the cases that trip people up.

Why black rectangles fail

A PDF page is built from instructions: "write these characters here", "draw this shape there". When you draw a black rectangle over a name, you add one more instruction on top. The instruction that writes the name is still in the file.

The result looks perfect on screen and on paper. But select the area and copy it, and the name lands in your clipboard. Search the document, and the reader jumps straight to it. Text extraction tools, search engines and AI assistants read the hidden words as easily as the visible ones.

This is not a theoretical risk. In 2019, a court filing by Paul Manafort's lawyers in the United States contained "redacted" passages that journalists recovered simply by copying and pasting them. Similar slips have exposed security manuals, contracts and personal data in many countries.

Step by step: redacting a text PDF

1. Keep the original. Redaction is permanent by design. Work on a copy.
2. Open the file in a real redaction tool. With pdfanonym, drop it on the redaction page; it opens locally.
3. Find every occurrence. Type a name, a reference or a number in the search box and press Mark all. Searches ignore capitals and accents, so "Müller" also finds "MULLER". Repeat for each piece of information.
4. Mark the rest by hand. Drag across passages, click single words, double-click a line to take all of it.
5. Apply. The text under each mark is deleted and replaced by a black box (or white space if you prefer).
6. Remove the metadata too. Leave the option ticked: the author, title and software fields often contain exactly the names you just redacted.
7. Download and check (see below).

Redacting a scanned PDF

A scan is a photo of a page. There is no text to delete, only pixels. Two consequences:

• Searching for a name finds nothing, so you have to locate each occurrence by eye.
• The redaction must change the image itself. In pdfanonym, switch to Area mode, tick images in the options and draw a rectangle over each passage: the pixels inside are blanked out of the image.

Some scans also carry an invisible OCR text layer, added by the scanner so the document can be searched. In that case the hidden text layer must go too. Area mode removes both the text and the image under the rectangle.

Five checks before you send

1. Copy test. Select the redacted area, copy, paste into a plain text editor. Nothing should appear.
2. Search test. Search the PDF for each redacted word. Zero results.
3. Properties test. In your reader, open File, then Properties. The author, title and subject should not reveal anything.
4. Comments test. Open the comments panel: review notes can quote the very passage you removed.
5. Attachments test. Look for a paperclip icon or an attachments panel. An embedded file can hold the unredacted original.

Redaction and GDPR

Under the GDPR, when you disclose a document in response to an access request, data about other people generally has to be removed or the disclosure has to be justified. Redaction is the usual way to do it. Doing it with an upload-based service means sending those people's data to yet another processor, which you would then have to account for. A tool that works locally avoids the question entirely.