How to Redact a PDF So the Hidden Text Is Really Gone

Every so often a court filing or official report makes the news because the “blacked out” passages could be read by anyone who selected the text and pasted it into a new document. The black bars were there, but the words underneath were too. Redacting a PDF properly is not difficult, but it does require understanding why the obvious approach fails.

Why a black box is not a redaction

A PDF page is built from layers of instructions: draw this text here, place this image there, paint a rectangle on top. When you add a black shape with an annotation or drawing tool, you are adding one more instruction to the stack. The original text is still in the file, still in the same position, still searchable and copyable. The box only hides it from people who look at the page the way you intended.

There are several easy ways to see through such a cover-up:

• Select the area with the text tool and copy it, then paste into any editor.
• Use the reader’s search function; it will happily find and highlight words under the box.
• Open the file in an editor that lets you move or delete annotations and drawn shapes.
• Convert the PDF to Word or plain text, which ignores the shape and extracts the text.

Highlighting with a dark color, changing the text color to white, or setting a tiny font size all fail for the same reason. So does placing a black image on top. And scanning a page you blacked out with a marker only works if the marker was truly opaque; some inks become see-through when the contrast is adjusted.

Other places sensitive data can hide

Even when the visible page is clean, a PDF can carry information elsewhere:

• Document properties and metadata: author name, company, original file name, software used, and sometimes a title or subject that reveals more than intended.
• Bookmarks and outlines: a bookmark titled with a client’s full name points straight at the section you blacked out.
• Comments and annotations: review notes, sticky notes and replies are stored separately from the page content.
• Form fields: the value in a field can remain in the file even if something is drawn over it.
• Attachments: a PDF can contain embedded files, such as the original spreadsheet behind a table.
• Hidden layers: some documents have optional content layers that are switched off by default but still present.
• The file name: easy to forget, and often the most revealing part of all.

How true redaction works

Real redaction removes the underlying content, not just its appearance. Specialized software does this in one of two ways: it edits the drawing instructions to delete the text and graphics inside the marked area, or it turns the affected page into an image with the area burned in, then throws the original page content away.

The vupdf Redact PDF tool takes the second approach because it guarantees nothing under a mark survives. You drag over areas to mark them, choose Black Out or Blur, and the tool renders each marked page as an image with the marked areas filled in. The page’s old text, fonts, images and annotations are removed, along with the form fields and bookmarks that belonged to those pages. Document-wide extras go too: attached files, the document properties (title, author, subject, keywords), the XMP metadata and document scripts. Leftover objects that nothing refers to anymore are dropped from the saved file. Pages without marks are left untouched.

There are trade-offs worth knowing. A redacted page becomes a picture, so its remaining text can no longer be selected or searched, and the file may grow somewhat. If you need the rest of that page searchable again, you can run OCR on the redacted result afterward. The tool also has an option to apply a marked area at the same spot on every page, which is handy for a recurring header containing a name or account number.

Personal data under KVKK and GDPR

In Turkey the Personal Data Protection Law (KVKK) and in the EU the GDPR both expect organizations to share only the personal data that a purpose actually requires. Redaction is one of the everyday tools for meeting that expectation when sending documents to customers, courts, auditors or the public. Typical items to check for include:

• Names, national ID numbers (such as the Turkish T.C. kimlik numarası), passport and tax numbers
• Home addresses, phone numbers and personal email addresses
• Bank account numbers and IBANs, card numbers, salary details
• Health information, such as diagnoses, prescriptions and sick notes
• Signatures, photos of people and license plates
• Information about third parties who are mentioned but not involved

Which data must be removed depends on the purpose and the recipient, and some categories, such as health data, are treated as especially sensitive under both laws. When in doubt about a legal obligation, ask your data protection officer or legal advisor.

A checklist before you share

1. Work on a copy and keep the unredacted original stored securely and separately.
2. Read the whole document once, looking for every occurrence of each item, including in headers, footers, tables, footnotes and appendices.
3. Mark every area, then use the redaction tool to produce a new file.
4. Open the result and try to select and copy text in the redacted areas. Nothing should come out.
5. Search the result for names, numbers and keywords you removed. There should be no hits.
6. Check the document properties and edit or clear anything revealing, and confirm there are no leftover comments, attachments or bookmarks that give the information away.
7. Rename the file if the name contains personal data.
8. Only then send or publish it.

The vupdf Redact tool does not search for terms automatically, so the reading step is yours. It is also worth doing a final visual pass at normal zoom: a mark that is a few pixels too small can leave the top of a line or the tail of a number exposed.

Bottom line

Drawing over text hides it from view but leaves it in the file. True redaction removes the content itself, and a careful check afterward confirms it is gone. Take the extra few minutes to look beyond the visible page, at metadata, bookmarks, attachments and the file name, because the information you meant to protect only has to leak through one of them.