Redaction failures are among the most reliably embarrassing mistakes in professional document handling, and they keep happening to organisations that should know better — law firms, government departments, courts, intelligence agencies. In nearly every case the cause is identical: someone drew a black box over text and believed the text was gone.
It was not. It was behind the box, fully intact, and anyone who selected the page and pressed Ctrl+C read it.
Why a black box is not redaction
A PDF is a stack of drawing instructions executed in order. Text is drawn, then your black rectangle is drawn on top of it. The rectangle is later in the sequence, so it appears above — but the text instruction is still in the file, unchanged.
Which means the original content survives every one of these:
- Selecting the page text and copying it into a text editor
- Running any text-extraction tool over the file
- Opening it in an editor and deleting or dragging the rectangle
- Searching the document for a term you thought you removed
- Reading the raw content streams, which requires no special skill
Real redaction is a different operation. It removes the text objects from the content stream, deletes the associated glyph data, and then draws the box. Nothing survives underneath because there is nothing underneath. This is a two-stage process in every tool that does it properly: you mark regions, then you apply, and the application warns you that content is about to be permanently destroyed.
If your tool did not warn you, it did not redact.
Tools that genuinely redact
Adobe Acrobat Pro has the most mature implementation. Its Redact tool marks regions, applies them destructively, and offers "Sanitise Document" to strip metadata and hidden content at the same time. It can also search the document for patterns — every instance of a name, or anything shaped like a national insurance number — and mark them all at once, which is what you want on a long document.
PDF-XChange Editor (Windows) includes redaction in its paid tier.
macOS Preview added a redaction tool in recent versions (Tools → Redact). It removes the underlying content rather than covering it. For occasional single-document work on a Mac, this is a genuine, free option that many people do not know exists.
Command line. mutool and qpdf can be scripted for bulk work, though redaction specifically is fiddly and worth verifying carefully.
Dedicated services exist for legal discovery and FOI work, with audit trails and review workflows. Appropriate when redaction is a compliance obligation rather than a one-off task.
The flatten-to-image method
Without access to any of the above, this works and is genuinely secure when done correctly — because it destroys all the text, not just the sensitive parts:
- Cover the sensitive regions with opaque black shapes.
- Convert every page of the PDF to an image — PNG or JPEG at 200–300 DPI. See how to convert a PDF to an image.
- Assemble those images back into a PDF.
The result contains only pictures of pages. There are no text objects at all, so there is nothing to extract from beneath your black boxes.
Be clear about what this costs: the document is no longer searchable, its text cannot be selected or copied, screen readers cannot read it, and the file will be considerably larger. For a three-page document you need to release safely, that is an acceptable trade. For a two-hundred-page report, it is not.
A caution on our own tool, to be accurate: Docento.app provides text, signature, and stamp annotations. It has no shape tool and no filled boxes, and text boxes are transparent — so it cannot produce the opaque black covers this method needs, and it does not redact. If you have seen advice suggesting a text box with a black background, that feature does not exist here. For redaction, use Preview, Acrobat, or a dedicated tool.
What people forget to redact
Removing the visible text is only part of the job. Documents leak through several other channels:
Metadata. Author name, organisation, the software used, creation and modification timestamps, and sometimes the original filename — which has by itself revealed authorship and internal case numbers. Acrobat's Sanitise Document handles this; exiftool does it free.
Hidden layers and objects. Content on hidden layers, cropped-away image areas, and objects positioned outside the visible page all remain in the file. Cropping a page hides content; it does not delete it.
Attachments and embedded files. PDFs can carry other files inside them, entirely invisibly.
Bookmarks, comments, and annotation history. Section titles in the bookmark tree routinely name things the body text was carefully edited to omit.
The text layer of a scan. A scanned document that has been OCR'd has an invisible text layer beneath the image. Blacking out the picture does nothing to it.
Revision history. PDFs can be saved incrementally, retaining earlier versions inside the same file. The pre-redaction version can sit in the file alongside the redacted one.
Redaction by inference. Black boxes reveal length. A short box after "Dear" narrows the name considerably. In documents where the same term is redacted inconsistently — caught once, missed once — the miss defeats every other box.
Verify before you release
Never send a redacted document without checking it. Three minutes:
- Select all the text and paste it into a plain text editor. If any redacted content appears, you covered rather than redacted. This single check catches most failures.
- Search the document for the names, numbers, and terms you removed.
- Inspect the metadata with
exiftoolor your reader's document properties panel. - Check the file size. If a heavily redacted document is the same size as the original, nothing was removed.
- Have someone else look, for anything consequential. The person who did the redaction is the worst-placed to spot what they missed.
The short version
Covering text hides it from your eyes and nobody else's. Real redaction removes the content and then covers the space, and only a tool built to do that can do it. If you are handling legal, medical, financial, or government material, use dedicated software and verify the output — the cost of getting this wrong is not a corrected document but a disclosed one.
Related: how to share a PDF securely, hidden data in PDFs explained, and how to password protect a PDF.
