How to redact a PDF — black out text vs. truly removing it

Redaction has a special place among PDF tasks: it is the one where getting it almost right is worse than not doing it at all. A blurry signature or a misaligned page is an annoyance. A redaction that only looks complete can expose exactly the information you meant to protect — to anyone who receives the file.

This guide explains the difference between covering text and truly removing it, walks through the classic way redactions fail, shows how to redact text locally in your browser with pdfmend, and is honest about when you should reach for a heavier tool instead.

A PDF page is not a flat picture. It is a set of instructions: draw this text here, this image there, this rectangle on top. When you draw a black box over a name, you are adding one more instruction on top of the others. The name is still in the file, underneath — selectable, searchable, and extractable.

This is why the distinction matters so much:

• Covering hides content visually. The underlying text remains in the document's data and can be recovered by anyone with the file.
• Redacting removes the content itself from the document's data. After true redaction, the text is not hidden — it is gone.

pdfmend keeps these as two clearly separate tools. Cover text (experimental) removes the original text operators from the PDF when possible, then draws a white cover and your replacement text — a cover-replace patch that suits visual corrections, but should still be treated as visual, never as certified redaction. Delete text (experimental) is the content-removal tool.

There is a well-known failure mode that has embarrassed organizations of every kind: a document is "redacted" with black rectangles, published or sent, and then a recipient simply selects the blacked-out region, copies it, and pastes the hidden text into a text editor — fully intact. Search also finds it, and text-extraction tools list it, because the text was always there. Only the pixels were covered.

No named incident is needed to make the point; the mistake is common enough to be a category. It happens because the visual result of covering and redacting is identical — a black box — while the file contents are entirely different. You cannot judge a redaction by looking at the page. You have to test the file itself.

pdfmend's Delete text tool removes text content from the document — on simple PDFs — and its results are verified on export. Here is the workflow:

1. Open the editor and load your PDF. Everything happens in your browser; a document you are redacting never touches a server.
2. Switch to the Delete text (exp.) tool and click the text line you need removed.
3. Press Delete text. Note that the tool is marked experimental: it removes matching text from simple PDF content streams, and it fails safely — declining rather than producing a false redaction — on PDFs with form fields, optional content layers, or rotated or duplicated pages.
4. Click Export PDF. The redaction is verified on export against the generated file.
5. Verify it yourself as well — this step is not optional. In the exported file, use Find in PDF to search for the removed text, and try selecting and copying across the redacted area. The text should be nowhere.

That last step deserves repeating: always test the exported file, the one you will actually send, not the document as it looks in any editor.

Honesty is the whole point of redaction, so here are the limits plainly. The Delete text tool works on selectable text in simple PDFs. It does not redact images or scans — if a document is a scanned picture of a page, the "text" is pixels, and pdfmend does not do OCR or image redaction. It declines files with form fields, optional content layers, and rotated or duplicated pages rather than risking an incomplete result. And no consumer tool should be your only line of defense when the stakes are legal or regulatory.

For those cases, two stronger options:

• Delete the whole page. If the sensitive content is confined to one page, removing the page in the pages sidebar is blunt but categorical — content on a deleted page does not travel with the export. You can also use Extract page to export only the pages you intend to share.
• Use a dedicated certified workflow. For court filings, regulatory disclosures, or anything where a failed redaction has legal consequences, use purpose-built redaction software with documented guarantees, and have the output reviewed.

Whatever tool you use, apply the same test: search the final file for the redacted text, and try to copy it out. For more on why local processing suits sensitive documents, see the local editing guide.