Redact a PDF
Permanently remove text from a PDF — not a black box drawn over it.
Drop your file here
Processed securely and deleted within the hour.
A black rectangle is not a redaction
This is the whole reason the tool exists. Drawing a filled black box over a name in a PDF hides it from your eyes and changes nothing else: the text is still in the file, still selectable, still copyable, and still returned by any extraction tool in about a second. Every reader can see straight through it.
Organisations with legal departments have published documents redacted that way, repeatedly — court filings, government reports, contracts. It is not an obscure failure. It is the default outcome of using the highlighter tool in a PDF reader and assuming it did what it looked like it did.
What this actually does
Pages containing something to remove are rebuilt as images. The page is rendered, the areas you specified are painted over, and that picture becomes the page. The result contains no text objects at all, so there is nothing left underneath to select, copy or extract — not the redacted words, and not the rest of that page's text either.
Pages with nothing to redact are copied through completely untouched and keep their selectable text, their file size and their quality. Most documents have one or two sensitive pages, so most documents come back mostly intact.
Why not just delete the words?
Because doing that correctly is much harder than it appears, and doing it incorrectly fails silently.
Removing individual glyphs from a PDF means walking the content stream, tracking the text and transformation matrices, resolving each font's width table, recursing into embedded form objects, and then inserting a compensating displacement so the rest of the line does not slide left. The tempting shortcut — ask a rendering engine which characters fall in the rectangle and use its indexes — does not work, and we tested it rather than assuming: a three-line page reports 50 glyphs in the content stream and 54 characters from the renderer, because the renderer inserts line breaks; a monospaced table reports 45 against 31, because it collapses runs of spaces. Code that indexes one list by the other's offsets deletes the wrong glyphs and reports success.
Rasterising is provably complete. That is worth more here than preserving selectable text on a page you are redacting.
It is verified, and it refuses rather than guessing
Before the file is handed back it is decompressed — every stream expanded, object streams disabled so nothing can hide inside a compressed container — and the raw bytes are searched for the text you asked to remove, in every encoding a PDF can hold it in: Latin-1, UTF-16, and hexadecimal. Then extraction is run over each redacted area and must come back empty.
If either check finds anything, you get an error and no file. A tool in this category should never hand back something that looks safe when it is not.
Everything else that hides information
Redaction is not only about the visible page. Also removed, document-wide: XMP and document-properties metadata, embedded file attachments, JavaScript, open and automatic actions, form fields — which store their values as text regardless of what is displayed — bookmarks, annotations and comments, and page thumbnails, which are small pictures of the page as it was before you changed anything.
Scanned documents
If your PDF is a photograph of paper, there is no text layer to search, and searching for a name will find nothing. Either run it through OCR first so there is something to find, or use the area option to cover a region by position — which is also how you redact a signature, a photograph or a stamp.
What we cannot promise
This removes what you tell it to remove and verifies that it is gone. It cannot know what else in your document is sensitive, it cannot judge whether your redaction is legally adequate, and it is not a substitute for reviewing the result yourself. Check the file before you send it. If you are redacting for a court filing or a disclosure request, your obligations are defined by that process and not by any tool.
The file is processed on our server, in a directory of its own, and deleted as soon as your download has been sent.
Common questions
Why does my redacted page look like an image now?
Because it is one. That is what makes the removal complete — a page with no text objects has nothing left to extract. Pages you did not redact are untouched and keep their text.
Can the text be recovered?
Not from the redacted pages: there is nothing to recover, and the finished file is decompressed and searched to confirm it. If that check fails you get an error instead of a file.
My PDF is a scan and it found nothing.
A scan has no text layer to search. Run it through OCR first, or use the area option to cover the region by position — which is what you need for signatures and photographs anyway.
Does it remove metadata too?
Yes — XMP and document properties, attachments, JavaScript, form fields, annotations, bookmarks and page thumbnails. All of those can carry content that is not visible on the page.
Will the file get bigger?
The redacted pages usually do, because an image of a page is larger than the text that drew it. You can lower the DPI, or compress the result afterwards.
How do I check a document somebody else redacted?
Use the redaction checker. It looks for text underneath black boxes and tells you plainly whether it is really gone.