Seven places a redacted PDF can still leak
A PDF is not a picture of a page. It is a container: the drawing instructions you see, the text those instructions place, and several other structures that most viewers never show. Each of them can hold the very data you meant to remove.
Text under the box
A shape drawn on top hides the text on screen; the characters stay in the page's content stream.
Check: Select all, copy, paste into a text editor. Search the PDF with Ctrl+F.
Invisible OCR layer
Scanned PDFs often carry a hidden text layer added by the scanner's OCR. Boxing the image leaves that layer intact.
Check: Run the same copy test on scanned pages — they should not paste the redacted words.
Document metadata
Title, Author, Subject and Keywords — and a second copy in XMP — often hold a client name or a case reference.
Check: Open Document Properties in your viewer (Inspector in Preview on a Mac).
Earlier revisions
An incremental save appends changes and keeps the previous version inside the file. The unredacted page can still be in there.
Check: Open the file in a text editor and count the %%EOF markers. More than one usually means earlier versions are kept (two can also mean a file saved for fast web view).
Hidden layers
Optional content groups can be switched off in the viewer but still be part of the file.
Check: Open the Layers panel of a desktop PDF reader and turn every layer on.
Comments and form fields
Reviewer names on annotations, sticky notes under the redacted area, values typed into form fields.
Check: Open the Comments panel; click into form fields to see their values.
Attachments
Embedded files travel with the PDF — the spreadsheet behind a table, the original e-mail.
Check: Open the Attachments panel (paperclip icon) of your reader.
Why the box alone is not a redaction
Almost every failed redaction has the same cause: something was added to the file, and nothing was taken out. A drawn rectangle, a black highlight or a darker text colour all add a layer on top. An incremental save adds the redacted version after the original instead of replacing it. In every case the sensitive data is still physically in the file.
It keeps happening in public
Court filings, government reports and corporate disclosures have been published with black boxes over names and figures that readers recovered with a simple copy-paste. The documents looked redacted on screen; the text layer had never been touched.
What a permanent redaction actually does
There are two sound ways to make a redaction permanent. One edits the file: the text objects under each box are deleted, the box is painted in, and the whole file is rewritten rather than appended to. The other rebuilds it: each page is turned into a new one with the boxes burned in, and nothing from the original is carried over by default.
RedactOffline rebuilds. When you export, in your browser:
- ✓Each page is rendered to an image, and every redaction box is painted into its pixels.
- ✓A brand-new PDF is assembled from those images — the original file is never edited.
- ✓Words outside your boxes are written back as an invisible layer so they stay selectable. A word a box touches is left out entirely.
- ✓Scanned pages get no text layer back, so an old OCR layer cannot carry redacted words through.
- ✓XMP metadata, attachments, scripts, bookmarks, links, hidden layers and earlier revisions are not copied.
- ✓Title, Author, Subject, Creator and Keywords are shown to you in the export dialog and kept only if you leave them filled in.
The trade-off is that pages become images: links and fillable form fields are not recreated, and text in non-Latin scripts is not selectable in the output. The details, field by field, are in how to remove metadata from a PDF.
A five-minute checklist before you send
Whatever tool you used, check the exported file — not the one you worked on.
Copy and search
Select all, copy, paste into a text editor; then Ctrl+F for a redacted word. Neither should find anything.
Extract the text
If you have it, run pdftotext (part of the free Poppler tools) on the file and search the output.
Document Properties
Read every field. A name or case number in Title or Keywords undoes the redaction on the page.
Panels you never open
Comments, Attachments and Layers: each should be empty, or contain nothing you redacted.
Earlier revisions
Open the file in a text editor and search for %%EOF. One occurrence is what you want; more usually means appended revisions.
Just want the quick version for the text under the box? See the 10-second copy test.
Export a PDF with nothing left behind
Every export is rebuilt from scratch in your browser. Nothing is uploaded. Free plan up to 5 pages.
Start Redacting For Free