Drawing a black rectangle over text hides nothing — the words are still in the file, one copy-paste away. Here's how real redaction works, what metadata gives away, and a checklist before you share.
Redaction failures are a recurring embarrassment: documents released with neat black boxes over the sensitive parts, from which anyone can recover the hidden text in seconds. No forensics involved — select-all, copy, paste into a text editor, and there it is. If you're going to redact a PDF, it's worth understanding why the obvious approach fails and what proper redaction actually does.
A PDF page is built in layers. The text lives in the page's content stream; shapes, highlights, and annotations sit in layers on top of it. When you draw a black rectangle in an editor, you're adding a new object above the text — the text itself is untouched underneath. Which means it survives, and it's trivially recoverable:
The same trap applies to white boxes over text, and to "highlighting" text in black. If the tool you used calls itself a drawing or annotation tool, you haven't redacted anything — you've decorated it.
The usual suspects are identifiers and money: ID and passport numbers, bank account and card details, home addresses, dates of birth, salary figures, names of third parties who never agreed to be in the document, and internal reference numbers that connect an anonymised file back to a person. Notice that most of these appear in predictable places — headers, footers, signature blocks, tables — and often repeat on every page, which is exactly why searching before you redact matters.
Real redaction removes the information instead of covering it. The text under the marked region is stripped out of the content stream, and the region is flattened into the page — typically by rasterizing it, so what remains is pixels with no letters behind them. After true redaction, there is nothing to recover: select-all finds nothing in the redacted area, text extraction returns nothing, and no editor can peel the box away, because there's nothing under it. That's the test that matters.
Even with the page content properly redacted, a PDF carries information around the page: title and author fields, XMP metadata, comments and annotations, bookmarks, embedded attachments, and sometimes earlier versions of the content left behind by incremental saves. An author name or an original filename in the document properties can undo an otherwise careful redaction — think "draft_offer_CompanyName.docx" sitting in the metadata of an anonymised bid.
So treat metadata removal as part of redaction, not an optional extra. After redacting the page content, strip the document's metadata and hidden extras — Arawa PDF's Remove PDF Metadata and Sanitize PDF tools handle exactly this.
Before the file leaves your hands, run through this:
Proper redaction takes a few minutes more than drawing boxes. Given that the failure mode is publishing exactly the information you meant to protect, those minutes are the cheapest insurance in document handling.