PDF Tools·4 min read

How to Redact a PDF So the Text Is Really Gone

Every few years a court filing or a government report makes the news for the wrong reason: someone copied the text from under the black bars and pasted it into a tweet. The bars were real, but the text underneath was still in the file. Proper redaction removes the content, not just the view of it. This guide explains the difference, walks through redacting a real document with the Redact PDF tool, and lists the checks that catch a leak before the file goes out.

Open the free Redact PDF All insights

Why a black rectangle doesn't redact anything

PDF pages are built in layers.

A PDF page is a stack of instructions: draw this text here, draw this image there, draw a rectangle on top. When you add a black box with a comment tool or a basic editor, you add one more instruction at the top of the stack. The text instructions underneath are untouched. Your eyes see black, but select the area and copy, and the words paste out cleanly. Search finds them, and so does any script that reads PDFs.

Real redaction removes the words from the file. Professional editors do this by editing the page's text instructions. The Redact PDF tool takes a simpler and very safe route: it draws each marked page as an image, paints the boxes into that image, and builds the new page from the image alone. The old text, fonts and drawings of that page are never copied, so there is nothing hidden left to find.

Three ways to mark what goes

Draw, search, or take the whole page.

Drawing is the direct way. Drag across the page to make a box, drag a box to move it, and drag its blue corner to resize it. With a keyboard, Tab to a box and use the arrow keys to move it, Shift with the arrows to resize it, and Delete to remove it. Boxes show see-through while you work, so you can check exactly what they cover.

Search finds text for you. Type words or phrases, one per line, and choose Match case or Whole words only if needed. Ready patterns find email addresses, phone numbers, card numbers, IBANs, US Social Security numbers, PAN and Aadhaar numbers, dates, web addresses and IP addresses. Every match is listed with its page and the words around it, and you untick the ones that should stay before marking the rest.

Whole pages suit annexures and attachments that shouldn't go out at all. Type a range such as 4, 9-11 and each page is covered completely.

A worked example: sharing a salary slip

Keep the proof, lose the personal details.

Say a landlord asks for proof of income, and you want to send your three-page salary slip PDF without your PAN, bank account and employee ID. Open the file in the tool. In Find text to redact, tick PAN numbers, then type your account number and employee ID on separate lines. Click Find, and the list shows every match: the PAN in the header of each page, the account number in the payment section, and the employee ID twice on page 1.

Click Mark to add boxes for all of them. Check page 2, where a deductions table repeats the employee ID in a narrow column, and drag the box a little wider so it covers the whole number. Pick black boxes and type REDACTED as the label. Leave Remove document properties ticked, because the file's author field holds your full name from the payroll system.

Click Apply redactions. The three marked pages are rebuilt as images, and the tool searches the new file again with your last search. The message says no matches remain. Download the file, open it in your reader, and try to select the area under a box. There's nothing to select.

What the redacted file contains

And what never makes it across.

Pages with at least one box become images at the resolution you pick, 150, 200 or 300 DPI, saved as PNG for sharp text or JPEG for a smaller file. Pages with no boxes are copied as they were, text and all, unless you tick Flatten every page. The document's title, author, dates and XMP metadata are removed by default. Bookmarks, attachments and scripts are not copied at all, because they can repeat the same details you just removed.

The trade-off is that redacted pages lose their selectable text. That's the price of certainty. If the rest of the page needs to be searchable, run the redacted file through OCR PDF afterwards. It can only read what's visible, so the redacted words stay gone.

Redaction mistakes that still leak

The box was fine. Something else wasn't.

Even with true redaction, details slip out in predictable places:

Repeated details. A name in the body is also in the header, the footer and the file name.
Scanned pages. Search can't see text in an image, so scanned pages need boxes drawn by hand, or OCR first.
Boxes that are a little too small. The tops of tall letters or a final digit can stay readable at the edge.
The file name itself, such as Rahul-Mehta-payslip.pdf. Rename the redacted copy before sending.
Sending the original. Keep the unredacted file in a separate folder so it can't be attached by mistake.

A two-minute check before you share

Trust, then verify.

Open the downloaded file in a different PDF reader from the one you normally use. Search for each name and number you removed. Page through every redacted page and look at the edges of each box. Try to select text on a redacted page, which should select nothing. Check the document properties in the reader and confirm the author field is empty.

It takes a couple of minutes, and it is the difference between a redacted document and one that only looks redacted. Because the tool runs in your browser, the unredacted file never leaves your device at any point in this process.