Guides / A black box is not redaction
Why a black box in Preview or Acrobat does not redact a PDF
Updated August 30, 2026
You drew a black rectangle over the account number. The page looks right. You attached it to the email. And the recipient can read the account number by selecting the page and pressing copy.
This is not a bug in your PDF reader. It is the difference between covering something and removing it, and almost nothing in a normal PDF toolbar removes anything.
What that rectangle actually is
A PDF page is a list of drawing instructions plus a separate list of annotations: comments, highlights, stamps, shapes. Markup tools add to the second list. The text you covered stays in the first, exactly as it was, because the two lists never touch.
That leaves two ways to get the words back, and neither takes any expertise:
- Select and copy. Text extraction reads the content stream. It does not care what is painted on top, so it returns the covered words along with the rest of the page.
- Move the box. An annotation is an object. In plenty of free readers you can click it, drag it aside, or delete it, and the text is simply there.
This has happened in filed court documents and in published government reports more than once, done by people whose entire job was to get it right. It is not a trap for the careless. It is a trap for anyone who has not been told.
What each tool really does
| What you did | What happened to the text |
|---|---|
| Preview markup rectangle | Untouched. A shape is drawn over it, and the shape can be selected and deleted. |
| Preview Redact (Ventura and later) | Meant to be removed. This is a different button from the rectangle, in the same toolbar. Verify the saved file. |
| Acrobat Reader comment or drawing | Untouched. Reader has no redaction feature at all. |
| Acrobat Pro, Mark for Redaction then Apply | Deleted from the page. This is real redaction, and Sanitize Document clears metadata too. |
| Black highlight in Word, then export | Untouched. Highlighting is formatting, and the exported PDF carries the characters. |
| Black-filled shape in any editor | Untouched, and often still a movable object in the output. |
| Export or print every page as an image | Gone, because there is no longer any text anywhere in the file. |
The test that settles it, in 20 seconds
Open the file you are about to send. Select all, copy, and paste into any text editor. If one covered word appears, you have not redacted anything. Searching the document for a name you removed does the same job. If you have poppler installed, pdftotext out.pdf - shows you exactly what a machine can lift out of the file.
Run this on the output every time, including when the tool told you it redacted the document. The failure is silent by nature: a page with the text still under the box looks identical to a page without it.
Scanned pages are not automatically safe
People assume a scan has nothing to extract. Two things spoil that. Scanners and phone apps routinely add an invisible OCR text layer so the document is searchable, and that layer copies out like any other text. And a box drawn over a scan is still an annotation sitting above the image, so deleting it exposes the pixels underneath. If the document matters, flatten it rather than trusting that it was never text.
The other places the words hide
- Document properties. Title, author and subject often carry a client name or a matter number.
- Attachments and comments. A PDF can carry embedded files, and reviewer comments quote the passage you removed.
- Form fields. A filled field keeps its value even when a box covers the box.
The blunt version that cannot fail quietly
FileQuiet's PDF redaction re-renders every page as a flat image with your boxes burned into the pixels, then re-opens its own output and counts the text objects left in it. That count has to be zero or no download is offered. Everything happens inside your browser tab, so the document you are redacting never travels, which is the one property most online redaction sites cannot offer.
Redact a PDF in your browserHonest limits: flattening removes all the text, not just the sensitive parts, so the result is not searchable or screen-reader friendly. If you have Acrobat Pro, its redaction keeps the rest of the document live and is the better tool for long reports.
Frequently asked
I flattened the file and the text is still copyable. Why?
Most viewers use the word flatten to mean merging form fields and annotations into the page. That bakes your box in permanently, which feels like progress, but it does nothing to the text drawing instructions underneath. Flattening in the sense that helps means re-rendering each page as an image.
Is a white box safer than a black one?
No, and it is worse in one way. It hides the fact that anything was removed, so nobody downstream knows to check. The extraction behaviour is identical: the words are still in the file.
Does password protecting the PDF fix this?
No. A password controls who can open the file. Once it is open, and you are sending it to someone precisely so they can open it, the covered text extracts exactly as before. Permissions that claim to prevent copying are advisory and widely ignored by tools.
Can I check a PDF somebody sent me for failed redactions?
Yes, and it is the same test. Select all, copy, paste. If the document has visible black boxes and the pasted text reads straight through them, the redaction failed. That is worth knowing before you forward it on.