You need to send a contract to a counterparty, but it shows your passport and your personal account number. Or accounting asks for a statement where only the amount should stay and the names of other employees have to go. Sometimes it gets trickier: a scanned medical note is heading to an insurer, and the diagnosis has to come out while the stamp and signature stay. The most dangerous move in all of these cases is to open the PDF in a viewer, draw a black box over the sensitive spot, and hit send. The recipient drags the cursor across the text, pastes it into a notepad, and everything you tried to hide shows up under the "fill."
The reason is that a PDF stores text and graphics in layers. A black box lands on top as a separate object, and the letters underneath do not go anywhere. They are still in the file, visible to copy, search, and any parser. A highlighter swipe or a white patch behaves the same way: clean to the eye, fully intact in the data. Real redaction works differently. It cuts the characters and pieces of the image out of the page for good and leaves a solid fill in their place with nothing behind it. That is how the redaction tool works: you mark an area, the content under it physically disappears from the document, and it can no longer be recovered from the finished file.
How to redact data in a PDF, step by step
1. Open Redact PDF and upload the document that has the sensitive parts. 2. Draw a box around each spot you need to hide: passport number, series, full name, address, account details, signature. Run the box a little wide so it catches the tails of letters and digits. 3. Go through every page to the end. Data often repeats in the header, in appendices, and on the back, not only on the first sheet. 4. Run the job and wait for the finished file. The marked areas get cut out of the content rather than painted over. 5. Download the result and check it the way the recipient will: try to select and copy the text in the redacted zone. 6. Keep the original separately. Redaction is permanent, and you cannot bring the original characters back from the finished copy.
What can go wrong
- A box instead of a redaction. Draw a rectangle in a regular editor and the text stays in the layer below it. Use the redaction tool, which deletes the content.
- Text peeking out from under the fill. A tight box leaves the tops and tails of characters visible, and a number or surname can be read from them. Cover an area wider than the text itself.
- Data left on other pages. Details repeat in headers, footers, and appendices. Page through the whole file, not just the sheet with the main text.
- A scan with recognized text. If an OCR layer sits over the image, redact the area so the recognized text disappears too, not only the visible part of the scan.
- A sensitive name in the metadata. The author name or original file name stays in the document properties. Clear them with Set PDF metadata.
- A password-protected file that will not open for editing. Remove the restriction first with Unlock PDF, then redact.
What to check before sending
- In the redacted zones the text does not select and does not copy, so the data is genuinely gone.
- Every page has been reviewed, including appendices, backs, and headers.
- Not a single character shows under the fill: no first digit, no last letter.
- The document properties hold no author name, original path, or comments with personal data.
- The original is saved somewhere separate in case the redacted copy needs reworking.
Related steps
Once the data is hidden, protect the file from accidental edits with a password using Protect PDF. If the document needs to be signed off before sending, add a signature in Sign PDF. And to make sure no old fragments resurfaced in the new version, compare it with the previous one in Compare PDF. Uploaded documents are deleted from the server after 120 minutes and are never used to train AI.