Redact PDF

How it works

Open the PDF, draw boxes over what you want to hide (or search for an e-mail, a phone number, an IBAN, an ID number or any text), see exactly what will be removed, and apply. The result is a new file in which the text, drawings and image pixels under each box are gone, not hidden behind a black rectangle.

The document information, XMP metadata, thumbnails, bookmarks, tagged structure, attachments, layer names and the value of form fields that no page shows, which can hold the same text, are cleaned as well, and you decide what to clean. Everything happens in your browser: the file is not uploaded to any server.

Redaction cannot be undone in the copy. Always keep the original and check the result before you share it: the page tells you plainly what was removed and what cannot be guaranteed.

What is not preserved, and limits

• Embedded fonts keep the shapes of the letters that were used (not their order) and each font's list of characters: the removed text cannot be read, but those technical leftovers stay in the file.
• A scan holds its text as pixels. The search cannot find it; only a box you draw removes it. There is no automatic detection of text inside pictures.
• Images we cannot decode exactly (JPEG 2000, JBIG2, CMYK JPEG) are removed whole, not just under the box. The result tells you how many.
• Patterns and soft-mask groups are not inspected: the result lists the pages that use them. Named destinations, page labels and the JavaScript in form fields are not cleaned.
• With fonts that have no width table we cannot tell which letter is under the box, so the whole piece of text that touches it is removed (more text than the box covers can go) and the text after it on that line can shift.
• A digital signature does not survive: the file is written again in full and the result carries no signature. The page warns you before you apply.
• Password-protected PDFs can be redacted (the page asks for the password and it is used only in your browser), but the result is saved WITHOUT a password. You can protect it again with Protect PDF.
• If a page has content we cannot read, it stops without changing anything: a page we could not read is never just covered.
• The result has been checked with open-source readers (poppler, pikepdf/qpdf, PDFium), not with Adobe Acrobat.
• Very large files (more than about 100-150 MB in total) may fail, especially on phones: everything is done in your device's memory.