How to Redact a PDF So the Text Is Actually Gone

The classic way to leak a document is to draw black rectangles over the sensitive bits and send the PDF out. The rectangle is a shape sitting on top of the page. The text is still underneath it, still selectable, still found by search, and anyone who drags a cursor across the black bar and pastes gets the name you were hiding. Real redaction has to destroy the text, not cover it.

Type the words instead of drawing over them

Upload the PDF and type what needs to go: a name, an account number, a street. Every occurrence is found across the whole document, blacked out, and the page is rebuilt so the words are no longer in the file. You never have to hunt for the fourteenth mention on page 31.

A few things about how the matching works, because they change what you should type:

• Separate the terms with commas. A term can be more than one word, so John Smith catches the whole name, as long as it sits on one line. Where a name breaks across a line ending, list the two halves as separate terms.
• Case doesn't matter. confidential catches CONFIDENTIAL and Confidential.
• Partial matches count. 2024 also blacks out 2024-06-15, and smith blacks out Smithson. Useful more often than not, but check the result if your term is short or common.

If none of the terms appear anywhere, the job stops with a message rather than handing you an unchanged file that looks redacted.

The document comes back as images

This is the trade-off that makes the redaction real. Rather than deleting characters and hoping nothing else in the file still references them, every page is rendered to a picture and the black boxes are painted onto the pixels. There is no text layer left to recover, from any page.

What that costs you:

• Search and selection stop working.Ctrl+F finds nothing, and you can't copy a paragraph out.
• The file usually gets bigger, sometimes a lot bigger, because page images take more room than text. Compress it afterwards if you need to email it.
• Document properties don't carry over.The output is a newly built PDF, so the original title, author and other metadata are gone. For a document you're about to hand over, that's usually the right outcome.

Check it before you send it

Two seconds of verification, every time. Open the redacted file, press Ctrl+F, and search for the word you removed. Then try to select the text under a black box. Finding nothing and selecting nothing is what a clean redaction looks like.

Worth thinking about separately: whether the redacted words can be inferred from what's left. A blacked-out name in a document that elsewhere lists the sender's email address hasn't been protected. Read the surrounding sentences as if you were the person trying to work it out.

Getting search back afterwards

If the redacted document still has to be searchable, run it through OCR once you're done. That reads the page images and adds a fresh text layer built from what's visible, so the words you removed can't come back. Do it in that order: redact, then OCR.

Common questions

What if the text I need to redact is in a scan?

There's no text to find, so nothing will match. OCR the scan first to add a text layer, then redact, then OCR again if you need search back. Check the result carefully: OCR that misreads a character means that occurrence won't be matched, and it'll still be sitting there in plain view.

Can I black out an area rather than specific words?

Not with this tool, which works from the words you type. When the goal is to remove a header, a footer or a signature block by position, cropping the margins is a better fit, keeping in mind that cropping hides content rather than deleting it. For something that must genuinely be gone, put a word from that area into the redaction list instead.