How to remove text from a PDF: what really disappears when you erase, cover or redact

Erase, cover, delete: what each tool actually does

Here is what really happens in the file depending on the action chosen, from the most common to the safest.

The last case calls for a clarification. In Adobe Acrobat Pro, the Redact tool works in two stages: you mark the areas, then you apply the redaction. Until it is applied, the marks are only annotations, and a file saved at that point keeps all its text. (In the French version of Acrobat, this tool is called "Biffer".)

Printing and then scanning the document does remove the text, since the result is an image; it is a heavy-handed solution that degrades quality and makes search impossible. As for online services that promise to delete text from a PDF for free, they may well erase correctly, but your file first passes through their servers, which raises a different confidentiality question.

Redaction failures that became famous

Redaction errors are not limited to individuals. Several high-profile cases rest on the same flaw: a black rectangle placed over text that was still there.

• January 2019, the Manafort case. In the United States, a filing by Paul Manafort's lawyers contained blacked-out passages. Journalists copied those passages into a word processor and read in clear information the defense had meant to withhold.
• December 2025, the Epstein files. Among the files released by the US Department of Justice, several media outlets showed that some blacked-out passages could be recovered with a simple copy and paste.
• Every day, in businesses. A contract sent with the salary "hidden", a report whose client name still appears in the file properties, a document whose file name contains the employee's name.

For an organization subject to the GDPR, these errors are not trivial. A redaction that leaves personal data readable may constitute a personal data breach within the meaning of article 4 of the regulation. Article 33 then requires notification to the supervisory authority (the CNIL in France) within 72 hours of becoming aware of it, unless the breach is unlikely to result in a risk to the people concerned. The European Data Protection Board details the process in its Guidelines 9/2022 on personal data breach notification.

What these cases have in common is reassuring: a simple check afterwards, searching for the hidden word or copying and pasting the blacked-out passage, would have caught the problem before release.

How to really remove text from a PDF: the method

A five-point method covers the essentials.

1. Work on a copy of the file, and keep the original separately.
2. Use a tool that deletes the text, and apply it all the way. A shape or a highlight is never enough.
3. Handle every occurrence: a name cited on page 2 often comes back on page 9, in a table or a note.
4. Clear the metadata (author, company, title) and check the file name.
5. Check the result: search for each removed value in the output PDF, then copy and paste the blacked-out areas into a text editor.

In a scanned PDF, the text does not exist as characters: it is part of the image. It must first be located by character recognition, then the corresponding pixels blacked out.

Biffer, the software we publish, applies this method on your computer, with no connection. It detects names, contact details, numbers and the fields of 18 professions, shows you a preview, then rewrites the text in place in the PDF: the value is replaced by an alias or a mask in all its occurrences, and the metadata are cleared. On a scan, character recognition runs locally and the pixels are blacked out permanently; the zone tool blacks out a chosen frame as is, such as a signature or a stamp. The same care applies to the original Word file, as the article anonymize a Word document explains. The page redact a document adds to this advice, and the free version is on the page download.