How to redact a PDF so the text is actually gone
Updated 19 August 2026
To redact a PDF properly, the sensitive text has to be removed from the page’s content, not painted over. UnboundPDF’s Redact tool deletes the matched text objects, draws the black box afterwards, and then re-reads the saved file to confirm the words are gone — telling you plainly when a page could not be fully verified. It runs in your browser; the document is not uploaded.
Covering text with a black box does not remove it. Here’s how to redact a PDF so the text is actually deleted — and how to check for yourself.
UnboundPDF is a free suite of 43 PDF and image tools that run entirely in your browser — merge, split, compress, edit text, OCR in 126 languages, redact, sign, convert and archive to PDF/A. Your document is read and written by the page on your own device; there is no document-upload endpoint in the core tools, no account, no watermark and no daily cap. Every result can be checked — with the Network tab, or with the Document Passport the Workspace writes for a chain of steps.
- Open Redact PDF and drop in one or more PDFs.
- Choose how to find the text: search a word or phrase, draw a box on a page, or pick built-in patterns (email, phone, IBAN, card numbers, US SSN, dates).
- Review the matches page by page. Untick anything that should stay.
- Optionally tick Sanitize to clear metadata, XMP, embedded files, scripts, annotations and bookmarks in the same pass.
- Apply, then read the result note — it says whether the saved file was verified clean, and names any page that could not be.
- Download. Open the file, try to select or search the redacted area: nothing should be there.
What a black box leaves behind
A PDF page is a stream of drawing instructions, and text is just one kind of instruction among them. When a tool “redacts” by drawing a black rectangle, it appends one more instruction — draw a box here — on top of the text instruction that is still in the stream. The words are still in the file. Select the area in a reader and you can copy them out; search the document and the term is found; run a command-line extractor and it prints them. Many leaked “redacted” documents in the news were exactly this: a cover, not a removal.
What proper redaction has to do
It has to delete the text instructions themselves and rewrite the page without them, then draw the box over the now-empty area so the layout still reads as redacted. UnboundPDF’s Redact PDF does this in your browser: it parses the page, removes the matched text operators, writes the page back, and then re-reads the saved bytes to confirm the words are absent — in plain, hex and two-byte encodings — before it tells you the file is clean.
Three ways to choose what goes
Search a word or phrase across every file you loaded, with whole-word and case-sensitive options. Draw a box on a page when you know where the sensitive area is. Patterns find things by shape: email addresses, international phone numbers, IBANs, credit-card numbers (checked with the card-number checksum), US Social Security numbers and dates. Tick more than one pattern and scan all your documents at once. Every match is listed per page, and nothing is removed until you apply — untick the matches that should stay.
When you see a black box instead of a deletion
Some content cannot be removed as text because it is not text: a scanned page, a logo with the number baked into the pixels, or a page whose structure the tool will not rewrite safely. For those pages the tool rebuilds the page as an image with the box burned in, and the result note names the page and says why. The same note uses the phrase NOT fully verified whenever the re-read could not confirm every page — for example when a term appears split across two drawing operators in a way that can be reported but not safely spliced. You are never told a file is clean when the tool could not check.
Sanitize: the rest of the file
Text on the page is not the only place a document remembers things. Metadata (author, title, custom fields), XMP packets, embedded files, JavaScript, annotations and bookmark titles can all carry the name or number you just redacted. Tick Sanitize and they are cleared in the same pass, and the re-read checks them too. If you redacted without it, Edit Metadata lets you inspect and clear the document properties afterwards.
How to check it yourself
Open the downloaded file in any reader. Select across the redacted area and try to copy — nothing should come out. Use the reader’s search for the term — it should find nothing. Try both on the original first so you know what a hit looks like. If you want to go further, the PDF Privacy Lab shows how to watch the network panel and confirm the document never left your browser while you worked.
What this does not promise
It does not promise that a scan has no handwriting in the margin you did not mark, or that a photograph does not show the number on a screen in the background — redaction finds what you search for or draw over. It does not rewrite what it cannot verify; it tells you. And it removes what you chose from the file you downloaded — the original on your disk is untouched, so keep or delete that copy deliberately.
Related tools
If the redacted document still carries form fields or comments, Flatten PDF bakes them into the page. To shrink it before sending, Compress PDF. If a margin or header just needs to be out of view rather than gone from the file, that is a different job — cropping a PDF without uploading it hides content, it does not remove it, so do not substitute one for the other. The shorter privacy-angle page is redact a PDF in your browser; the step-by-step for personal data is find and redact SSNs, IBANs and card numbers; the comparison of approaches is redaction that removes text vs a black box.
Frequently asked questions
Does redacting a PDF actually remove the text?
Only if the tool deletes the text from the page content. A black box drawn over text leaves the words in the file — selectable, searchable, extractable. UnboundPDF removes the text operators, then re-reads the saved file to confirm the words are gone, and says so when a page could not be verified.
How can I tell whether a redacted PDF is safe to send?
Open it and try to select and copy the redacted area, then search the document for the term. If either finds the text, it was covered, not removed. Do the same on the original first so you know what a hit looks like.
Can I redact the same number across several PDFs at once?
Yes. Load the files, search the term or tick a pattern such as IBAN or card numbers, review the per-page match list, and apply. Matches you untick are left alone.
Does redaction remove metadata too?
Tick Sanitize and the same pass clears metadata, XMP, embedded files, scripts, annotations and bookmark titles, and the re-read checks them. Without it, only the page content you chose is changed.
Why does the result say “NOT fully verified”?
Because the tool could not confirm every page is clean — a scanned page that had to be rebuilt as an image, or a term split across drawing operators that can be reported but not safely spliced. The note names the page and the reason so you can decide.