The Mistake That Keeps Leaking Secrets
Someone needs to share a document but hide a few sensitive lines — a name, a bank account number, a salary figure. So they draw a black rectangle over the text, save the PDF, and send it off believing the information is gone.
It isn't. In most cases the text sits untouched *underneath* the black box, fully intact and fully copyable. Anyone can select the "hidden" text and paste it somewhere, or delete the box and read what's beneath. This isn't a rare edge case — it's one of the most common and most damaging privacy mistakes made with documents, and it has burned law firms, courts, and government agencies repeatedly.
The core misunderstanding is simple but critical: covering text is not the same as removing it.
A Failure That Made Headlines
In January 2019, lawyers for Paul Manafort filed a court document as a PDF with black rectangles over the sensitive passages. Within hours, reporters simply copied the text out from under the boxes and pasted it into a plain document — revealing material the defense had fought to keep sealed, including that Manafort had shared polling data with a Russian associate. The "redaction" was defeated with Ctrl+C and Ctrl+V.
It's a perfect illustration because the people involved were sophisticated, well-resourced professionals with every incentive to get it right — and the black-box method still failed instantly. The same mistake has surfaced in government manuals, corporate filings, and public-records releases. The tool changes; the error is always the same.
Why the Black Box Fails
To understand the failure, you have to understand how a PDF is built. A PDF page is a stack of separate objects drawn in order — text, images, shapes — each with its own coordinates. When you draw a rectangle over some words, you add one new object on top of the stack: an opaque black shape.
The text object underneath is completely unaffected. It's still in the file, still at its coordinates, still recording exactly which characters it represents. The black box only changes what's painted on screen — it's a sticker over a word, not an eraser.
| Test | Fake redaction (black box) | Real redaction |
|---|---|---|
| Select & copy the area | Reveals the hidden text | Copies nothing |
| Delete the covering shape | Text reappears | Blank space |
| Search (Ctrl+F) for the words | Finds them | No matches |
| Inspect the raw file | Text still present as data | Text is gone |
The same trap applies to highlighter-style annotations, "white-out" boxes, and any markup tool that layers something over content rather than deleting the content itself. The leftover text is the same kind of invisible-but-present data we cover in PDF metadata and what's hidden in your documents.
What Real Redaction Actually Does
Proper redaction doesn't cover text — it destroys and replaces it. A true redaction tool:
1. Identifies the exact text and image data in the region you're redacting 2. Permanently deletes that underlying content from the file 3. Draws the black bar in its place as a visual indicator that something was removed
After genuine redaction, there is nothing under the black bar to recover. Selecting the area copies nothing. Deleting the bar reveals blank space. Searching finds nothing. The information is actually gone from the document, not merely hidden from view.
The visual result looks identical to the fake version — which is exactly why the fake version fools people. The difference is entirely in what happened to the data underneath.
"But I Flattened It" — Why That's Not Enough
A common follow-up mistake: people cover text with a box, then *flatten* the PDF, assuming flattening bakes everything down and destroys the hidden text. It usually doesn't.
Flattening merges interactive layers — form fields, annotations, comments — into the static page. But text sitting underneath a covering shape is ordinary page content, not an annotation, so flattening doesn't necessarily remove it. You can end up with a flattened file that still has the original words recoverable in its content stream.
The only reliable ways to eliminate the underlying data are true redaction (which deletes the specific content) or rasterizing the entire page into an image (which discards the text layer altogether — more on that below). Flattening alone is not redaction.
How to Redact a PDF Safely
Use a real redaction feature, not a drawing tool. Look specifically for a function labeled "Redact" (Adobe Acrobat has a dedicated Redaction tool; some other editors do too). If a tool only offers "draw rectangle," "highlight," or "add shape," it cannot truly redact — it can only cover.
Don't forget images. Redaction isn't only about text. Faces, signatures, handwritten notes, and barcodes in embedded images may also need removing — and a barcode or QR code can encode data that survives a box drawn over the human-readable text beside it.
Remove metadata too. Redaction handles the visible page, but sensitive information can also hide in the document's metadata, comments, and residual content. A document isn't fully sanitized until both the page content and the hidden data are cleaned.
Then verify — every time. This is the step people skip. After redacting:
- Try to select and copy the redacted area. You should get nothing.
- Use Ctrl+F to search for the redacted words. There should be no matches.
- If it's truly sensitive, check the raw file to confirm the text strings are gone.
If any of those tests turn up the "removed" text, the redaction failed and the document is not safe to share.
The Belt-and-Suspenders Method
For the highest-stakes documents — legal filings, medical records, anything where a leak has real consequences — professionals often add a step that removes all doubt:
Print and rescan (or rasterize). Convert the redacted document to a flat image — print to paper and scan it back, or export each page as an image. The result has no underlying text layer at all, so there is provably nothing to recover. The trade-off is that the file becomes an image-based PDF with no selectable text, the same limitation we describe in why your scanned PDF won't highlight. For a document you're redacting precisely *because* it's sensitive, losing the text layer is usually an acceptable price.
A Note on Where You Redact
If a document is sensitive enough to need redaction, think about where the redaction happens. Uploading a confidential file to an online tool means the *original, un-redacted* file — with every secret still in it — travels to and sits on someone else's server. You'd be exposing the exact information you're trying to protect, during the act of protecting it. This is the broader concern we cover in the hidden risks of uploading PDFs to free online tools: for truly private work, the safest redaction is one that never leaves your device.
Frequently Asked Questions
Can redacted text be recovered?
If it was covered with a black box, highlight, or white-out shape — yes, trivially, by selecting, searching, or deleting the shape. If it was removed with a true redaction tool or the page was rasterized into an image, then no, there's nothing left to recover.
Does flattening a PDF redact it?
No. Flattening merges annotations and form fields into the page but does not reliably delete text hidden under a covering shape. Use a real redaction feature or rasterize the page instead.
Is drawing a black rectangle ever safe?
Only if you also permanently delete the content beneath it — which is what a redaction tool does automatically and a drawing tool does not. On its own, a rectangle is just a visual cover, never a safe redaction.
The Bottom Line
A black box is a visual cover, not a delete key. Real redaction removes the underlying text and image data permanently, leaving nothing to select, search, or recover — as the Manafort filing showed, even experts get burned when they forget this. Use a dedicated redaction feature rather than a drawing tool, remember that flattening is not redaction, strip metadata alongside the visible content, and always verify by trying to select and search the redacted area. When the stakes are high, rasterizing the page removes any remaining doubt — and keeping the whole process on your own device ensures the unprotected original never leaves your hands.