PDF Redact
Black out sensitive PDF content for real: marked pages are rebuilt as flat images so the text underneath is removed, a self-check re-extracts text from the output to prove zero residue before download, and document metadata is wiped. 100% in your browser.
No file handy?
Marked pages are rebuilt as flat images — the text underneath is removed from the file, not hidden. Before the download appears, the tool re-opens its own output and proves that zero text remains on every redacted page. Document metadata (title, author…) is wiped too.
Drag on the page to draw a box over anything sensitive. Boxes snap to this page only.
Redaction that actually removes the text
Drawing a black rectangle over text in a normal PDF editor hides nothing: the words are still in the file and come right back out with copy-paste or a search index. This tool rebuilds every marked page as a flat image, so the text objects underneath no longer exist — then it re-opens the result and extracts text from each redacted page to prove the count is zero before offering the download.
Pages you didn't mark are copied through untouched, and the whole thing runs in your browser: the document you're redacting is exactly the kind of file you shouldn't upload anywhere.
Most "redaction" is fake: a black rectangle drawn over text hides pixels, not data. Select the covered area, copy, paste — and the secret is back. Court filings and government releases have leaked exactly this way. True redaction has to remove the text from the file itself, which is what this tool does: every page you mark is re-rendered as a flat image, so the words underneath simply no longer exist in the output.
The tool then proves it. After building the redacted PDF it re-opens its own output and runs text extraction on every redacted page — the download button only appears when the count comes back at exactly zero characters. Document metadata (title, author, subject, keywords, creator) is wiped in the same pass, because file properties leak names and organizations too.
A document that needs redaction is, almost by definition, a document you should not upload to a random website. Everything here — rendering, blacking out, rebuilding, self-checking — runs locally in your browser. Pages you didn't mark are copied through unchanged, with their text still selectable.
FAQ
Why does it rasterize pages instead of just drawing black boxes?
Because a black box on top of text is cosmetic: the text objects are still in the file and can be recovered with copy-paste, text extraction, or a search index. Rebuilding the marked page as a flat image removes the text layer entirely — there is nothing left to extract. That is also why text on redacted pages is no longer selectable.
What exactly does the self-check verify?
After generating the output, the tool re-opens it and runs text extraction on every redacted page. Only if zero text characters come back — and the page count matches the original — does the download button appear. If anything is extractable, no file is offered.
Do unmarked pages change?
No. Pages without boxes are copied through with their original content: text stays selectable, links keep working. Only marked pages are rebuilt as images.
What happens to the PDF metadata?
Title, author, subject, keywords, creator and producer are cleared in the output — document properties often leak names, employers and software versions. Remember the filename is still yours to choose.
Is my document uploaded?
No. Rendering, redaction, rebuilding and the self-check all run in your browser. A document that needs redaction is exactly the kind of file that should never touch a third-party server.
Does it work on scanned PDFs?
Yes — scanned pages are already images, so drawing a box over the sensitive area and rebuilding works the same way. Note this tool redacts what is visible; it does not run OCR.
Why did my file get bigger or smaller?
Redacted pages are re-encoded as JPEG images at your chosen resolution (96/144/216 dpi). Text-heavy pages usually grow a little; image-heavy pages often shrink. Unmarked pages are unaffected.