A Black Box Is Not Redaction

Why the rectangle does nothing

A PDF is a stack of drawing instructions executed in order. "Draw this text here" comes first; "draw a filled black rectangle here" comes later, so the rectangle covers the text on screen. Neither instruction knows about the other, and the text instruction is still in the file with its content intact.

Copy and paste through the rectangle and you get the text. Open the PDF in any text extractor, or in a browser's developer tools, and you get the text. Convert the PDF to Word and the text reappears, now with a black box floating uselessly beside it.

The other places information hides

Even properly removing visible text is not the whole job, because PDFs carry information that never appears on a page. This is how documents leak details nobody thought were in them.

• Metadata: author name, the organisation from the software licence, and the full path of the file on the creator's machine.
• Revision history: some PDFs carry earlier versions of the content within the same file.
• Embedded attachments: spreadsheets and images tucked into the document, invisible on the page.
• Cropped images: cropping in a PDF often hides part of an image rather than deleting it, so the cropped-off part is still in the file.
• Layers: hidden layers used in the original design, still present and extractable.

How to redact properly

Use a tool that removes the underlying content rather than drawing over it. Proper redaction deletes the text objects in the selected area and then draws the black box, so there is genuinely nothing left to recover.

For anything genuinely high-stakes, there is an older approach that cannot fail: print the redacted document, black out the sensitive parts with an opaque marker on paper, and scan the result. You lose searchable text and file size, and you gain absolute certainty, because a scan of a page contains only what was on the page.

Verify before you send

Never trust a redaction you have not tested. The test takes fifteen seconds and it is the difference between a redacted document and an incident.

• Export the file and reopen it — do not test in the editor.
• Select the redacted area with your cursor and press Ctrl+C. Paste into a text editor. Nothing should appear.
• Use the reader's own search to look for a name or number you removed. It should find nothing.
• Check the document properties for author, company and file paths.
• If you can, convert the PDF to plain text and read the whole output — that shows you everything a determined recipient would find.

When redaction is the wrong tool

If most of a document is sensitive, do not redact it — extract the parts you are willing to share into a new document. A page with fifteen black boxes is both harder to verify and an invitation to try recovering them.

And if you are sharing a document because someone needs a single figure from it, send the figure. The safest redaction is the page you never sent.