What it does
You open a PDF, drag rectangles over whatever must not be seen (an account number, a name, a signature, a whole paragraph) and press one button. The result is a new PDF in which those areas are solid black and the information underneath no longer exists anywhere in the file.
That last part is the point. Most "redaction" that people do, such as drawing a black rectangle in a PDF editor, a presentation app or a markup tool, only paints over the text. The words are still in the file, one copy-and-paste away. We explain exactly why in Why black-box PDF redaction fails.
How it works
- Render. pdf.js (bundled into this site) draws each page to a canvas so you can see it.
- Mark. You draw boxes with your mouse or finger. You can also type a word and the tool will box every text run that contains it. Boxes are stored as fractions of the page, so they stay exactly where you put them.
- Burn in. On export, each page is re-rendered at your chosen resolution into a fresh canvas, and the boxes are filled with opaque black on those pixels. At this moment the original pixels under each box are overwritten.
- Rebuild. A brand-new PDF is created with pdf-lib. Each page contains only a picture at the original page size. The original file, with its fonts, text, annotations, form fields, bookmarks and metadata, is never copied into the output.
- Self-check. We open the rebuilt file with pdf.js and ask it for the text on every page. You will see the count, which should be zero characters and zero fonts.
Why the text is genuinely gone
A normal PDF page is a small program: "use this font, put the characters 'Account 4417…' at this position, then draw a black rectangle at this other position". A black-box tool adds one more instruction; the earlier ones stay. Delete the rectangle, select the text underneath, or run a text extractor, and the secret appears.
Our output has no such instructions. Each page is one image (a grid of coloured pixels), and the pixels under your boxes are black. There is no font, no character, no text object to extract, and no hidden layer to peel back. You can prove this yourself: open the result in any viewer, press Ctrl/⌘ + F to search for a word you redacted, try to select text, or run pdftotext on it. You'll get nothing.
A worked example
You have a three-page invoice that you need to share with an accountant but without your bank details. You drop invoice.pdf into the tool and see page 1. You draw a box over the IBAN in the footer, type "IBAN" into the search field and press Mark matches to catch the two other places it appears, and then drag a box over a signature on page 3. The counter reads "4 boxes on 2 pages". At 200 dpi with JPEG pages you press Redact & rebuild PDF. After a few seconds, the green self-check reads "0 characters, 0 fonts, 3 page images". You download invoice-redacted.pdf, open it, search for "IBAN" and find nothing, because there are no words left to find, in the redacted places or anywhere else.
Limits you should know about
- The whole document becomes images. That is what makes redaction real, but it means the output is not searchable, text can't be selected, and screen readers can't read it. If you need an accessible or searchable copy, run OCR on the result separately, after redacting.
- File size changes. Page images are usually larger than the text-based original. 150 dpi JPEG gives small files; 300 dpi PNG gives large, very crisp ones.
- "Mark matches" boxes whole text runs. A PDF stores text in chunks, so if one chunk is "Total: IBAN DE89 …", the whole chunk is covered. That errs on the side of hiding too much. Always review the boxes before exporting. It can't find text inside scanned images, so draw those by hand.
- You are responsible for the review. The tool burns in what you marked. Anything you did not box remains visible.
- Everything else is dropped, including links, bookmarks, form fields, annotations, attachments, signatures and document metadata. That is deliberate.
- Encrypted PDFs can't be opened. Remove the password first.
- Memory. Rendering happens in your browser. A very long or very large-format PDF may be slow or fail on a phone; try fewer pages or a lower resolution.
Verify there are no uploads
You shouldn't have to take our word for it. This page is built so you can check in under a minute:
- Open your browser's developer tools (
F12, or⌥⌘Ion a Mac) and switch to the Network tab. - Tick Preserve log, then click the clear (🚫) button so the list is empty.
- Add your file and run the tool.
- Look at the list. You will see nothing new: no
POST, noPUT, no calls to any other domain. Your file is read from disk into memory and handed to code that was already loaded with this page. - For a stronger test, switch the Network tab's throttling to Offline (or turn on airplane mode) after the page has loaded, and use the tool again. It keeps working, because it never needed the network.
PDF.js starts its worker script when the page loads, so you will not see even that after you add a file. Every library is bundled into this site and served from the same domain. There are no CDNs, analytics scripts, fonts or trackers. Our full guide to verifying uploads covers Chrome, Firefox and Safari.
Frequently asked questions
Is this really permanent?
Yes. The redacted areas are overwritten with black pixels, and the output file is built from those pixels only. The original text objects, fonts and metadata are never written to the new file. The built-in self-check re-reads the output and reports the amount of extractable text.
Can someone recover the text by removing the black boxes?
No. The boxes are not separate objects sitting on top of the page. They are part of the page image. There is nothing to remove, and the original pixels beneath are gone.
Why is my output bigger than the original?
Text-based PDFs are compact; pictures of pages are not. Try 150 dpi and JPEG for smaller files.
Can I make the result searchable again?
Not here. If you need that, run an OCR tool on the redacted file afterwards. It will only see the words that remain visible.
Does it work on scanned PDFs?
Yes. You draw boxes by hand. "Mark matches" only works on PDFs that have a text layer.
Does it work on a phone?
Yes. Use the Draw / Scroll switch: Scroll lets you move the page with your finger; Draw lets you drag boxes.
What happens to the original file?
Nothing. We never modify it. The redacted version is a separate download. Keep the original somewhere safe, because it still contains everything.