How to redact a PDF so the text is actually gone

The mistake that keeps making headlines

Court filings, government releases, corporate documents — every year, redacted PDFs leak their secrets, and it is almost always the same mistake. Someone used a highlighter tool set to black, or drew a filled rectangle, over the sensitive text. The page looks redacted. But a PDF page is layers: the text is one layer, the rectangle another sitting on top. Select-all-and-copy grabs the text straight through the box. Search finds it. Screen readers speak it aloud. Removing the rectangle in an editor reveals it. The information was never gone — just decorated.

What real redaction means

True redaction rewrites the page content itself: the sensitive text is deleted from the file's internal structure — the characters cease to exist anywhere in the document — and only then is a black bar drawn where they used to be, as a visual marker of the removal. After real redaction, searching for the redacted words finds nothing, because there is nothing. Copying yields nothing. No editor can bring the words back, because the file no longer contains them. That destructive rewrite is the entire point, and it is what separates a redaction tool from a drawing tool. Our redaction tool works this way — removal first, black bar after.

The other places secrets hide

Removing the visible text is necessary but not always sufficient. A thorough redaction pass also considers:

• Scanned pages. On a scan, the “text” is part of the page image, and if OCR has been run there is also an invisible text layer behind it. Real redaction on a scan must black out the image region and strip the matching hidden text — covering one but not the other leaks.
• Metadata. The document's title, author and creation trail live outside the visible pages. A file named after the confidential project, authored by the person you're anonymising, undoes careful page-level work.
• Repeated occurrences. A name you redacted on page 2 probably also appears on pages 7, 15 and in a footer. Search-based redaction — find every occurrence, review the list, remove them all — beats scrolling and spotting by hand every time.

A review step that catches the classic errors

Good tools show you every match before removing anything, because context matters: redacting the surname “Stone” shouldn't silently delete “limestone” and “Stonebridge Road”. Review the match list, untick the false positives, then apply. After applying, run the checks that would have caught every famous redaction failure:

1. Search the document for the redacted term — zero results expected.
2. Select all → copy → paste into a text editor — the redacted words must not appear.
3. Check the file's properties for names in title/author fields.

Thirty seconds, and you know rather than hope.

Why “delete the page” and “password it” are not redaction

Two tempting shortcuts, both flawed. Deleting the whole page loses the ninety-five percent of it that should stay public. Password-protecting the file (see our protection guide) restricts who opens it — but everyone authorised to open it sees everything, and the password will be shared. Redaction is for when the document should circulate and specific contents should not. Different problems, different tools.

Last practical note: documents needing redaction are, by definition, sensitive — so prefer a tool that redacts on your own device rather than uploading the unredacted file (secrets intact) to someone's server. And always redact a copy; keep the intact original somewhere safe.

Keep reading