Redacting a PDF Properly: Why Black Boxes Are Not Enough

The most common redaction mistake is invisible. A black box drawn over a name looks perfect on screen and on paper, but in most cases the name is still in the file, directly underneath. Anyone can select the text under the box and paste it somewhere else, or find it with a simple search. Court filings, government reports and company documents have all been published this way, with the "hidden" text recovered in seconds. This guide explains why covering is not removing, the less obvious places information hides in a PDF, and how to check a redacted document before it leaves your hands.

Why a Black Box Hides Nothing

A PDF page is built up in layers, like a stack of transparent sheets. The text is drawn on one, and a black rectangle drawn later sits on a sheet above it. Your eyes see only the top of the stack, but every layer is still in the file. Copying, searching, text-to-speech software and search engines all read the text layer directly, and the rectangle makes no difference to them.

The same goes for most quick fixes:

Black highlighting or a filled shape added in a PDF viewer is a comment that sits on top of the page. Anyone can select it and delete it, or move it aside.
Black highlighting in Word stays as text in a black-on-black format. When the document is saved as a PDF, the words go with it.
Changing the text color to white hides it from view, and from nothing else.

The Less Obvious Places Information Hides

Even when the visible text is dealt with properly, a PDF can carry the same information in places nobody looks at.

The text layer of a scan. Many scanned PDFs have invisible OCR text behind the image, so that they can be searched. Covering the picture of a word leaves that invisible copy of it untouched.
The pixels of an image. A box drawn over a photo or a scanned signature covers it, but the image data underneath is complete. Removing the box, or extracting the image, shows the original.
Metadata. The document's title, author, subject and keywords, and the file name itself, often contain the very names being redacted.
Comments, form fields and bookmarks. A reviewer's note, a filled-in form value or a bookmark called "Settlement with J. Smith" can repeat what the page hides.
Attachments. A PDF can carry other files inside it, such as the spreadsheet the figures came from.
Earlier versions. Many PDF editors save changes by adding them to the end of the file, so the version before the redaction is still inside it.
Accessibility information. Tagged PDFs can carry a replacement text for screen readers that spells out words drawn on the page.

Saving the file is a trap of its own. When we tested one online redaction service while building our own tool, it removed the words from the page correctly. However, it saved the file with the page's original content still inside it, as an object that nothing referred to. No viewer displayed it and text extraction found nothing, yet the redacted values could be read straight out of the file. A redacted file is only as good as the way it is written.

What Proper Redaction Does

Real redaction changes what the file contains, not just how it looks:

the characters under each marked area are deleted from the page's content;
the covered pixels of any image are overwritten in the image itself;
drawings that the box hides completely are deleted;
a black box is then drawn in the empty space, as a visible marker that something was removed;
the metadata, comments, form fields, bookmarks, attachments and other hidden content are removed as well; and
the file is written out fresh, with no earlier versions and no left-over objects.

There is also the question of letters a box only partly covers. A box that clips half a digit can leave enough of it to read, so a careful tool takes a word the box clips out whole, rather than leaving half of a card number on the page.

How Redact PDF Works

Our Redact PDF tool is built around those rules. You open the PDF, then select text or draw boxes over anything that has to go, including parts of images. The file is uploaded to our servers, where it is processed automatically and deleted about five minutes after you close the page. The limit is 256 MB.

What a box covers is taken out of the file itself. The text is removed a character at a time, and a word that a box only partly covers may be removed whole. The covered pixels of a picture are blanked in the picture. The rest of the page is left exactly as it was. The tool also removes the document's metadata, comments, form fields, bookmarks, attached files and document scripts, along with the accessibility structure that could spell out removed words.

Then it checks its own work. It reads the finished file back, decompressing every piece of content in it, including any that nothing refers to, and looks for the words that were under your boxes. If one of them is still in the file, it tells you instead of handing over the file.

There are limits to what any tool can do for you. Redact PDF removes what you mark, and only that. If the same name appears on another page, in a header or inside a picture you did not mark, it stays. The check looks for words that existed only under your boxes, so it cannot know that a name elsewhere in the document also matters. And redaction cannot be undone, so keep your original somewhere safe.

How to Check a Redacted PDF Yourself

Whatever tool you use, check the result before you share it. It takes a few minutes:

Search for every redacted term. Use Ctrl+F (Cmd+F on a Mac) for each name, number and phrase you removed. Try part of it too, such as the last four digits of a number.
Copy everything. Select all the text on each redacted page, paste it into a plain text editor, and read what comes out.
Use two different readers. Open the file in a browser and in a desktop PDF reader. They read files differently, and one may show what the other does not.
Look at the document properties. Check the title, author, subject and keywords, and check the file name.
Open the side panels. Look at comments, bookmarks, attachments and form fields.
Read every page. Look for the same information somewhere you did not mark: a header, a footer, a caption, an image, a signature block.

Conclusion

Redaction is about what is in the file, not what is on the screen. A black box over text only hides it, while real redaction deletes the text, blanks the image pixels, clears out the hidden data and writes a clean file. Use a tool that removes content rather than covering it, mark every place the information appears, and check the result as if you were the person trying to read it.

Have a document to redact? Try Redact PDF.

← Back to the blog