Redact PDF
Blacked out means gone, not covered up
…
- 1
Open the document
Up to 50 pages. It is read from disk and never sent anywhere, which for this job is the whole point: a document worth redacting is a document worth not handing to a stranger first.
- 2
Mark what has to go
Drag over anything that should not be there. Or press the one button and have a reader go through the pages first, finding names, addresses, phone numbers, accounts, dates and passwords in the text and boxing them for you. Either way you decide what stays: every box comes off with a click.
- 3
Save it
The pages come back as pictures, which is what makes the removal permanent. There is no text layer left for a covered word to survive in.
FAQ
- Why is drawing a black rectangle not enough?
- Because in a PDF the rectangle and the text are two separate things sitting on top of each other. The words are still in the file, still selectable, and still there for anything that reads the document instead of looking at it. Real documents have leaked exactly this way, and the people who published them believed they had redacted them. Selecting the area and pressing copy is all it takes.
- What does this do instead?
- It draws every page into pixels, paints the boxes onto those pixels, and builds a new document out of the pictures. There is no text layer in the result at all, so there is nothing for a covered word to hide in. You can check it rather than take our word for it: open what you saved and try to select a line of it.
- What does that cost me?
- The document can no longer be searched, selected or copied from, by anyone, including you. It also gets bigger, because pictures of pages are heavier than the text on them. Keep the original somewhere safe and treat the redacted one as the copy you hand over.
- Is my document uploaded?
- No. Every page is drawn, marked and rebuilt on this page, on your own machine, and no request carries the document anywhere. This is the tool on this site where that matters most, which is why it works the way it does.
- How good is the automatic finding?
- Good enough to find the things a pattern cannot. It reads the text the way a person skims it, so it picks up a name because of the words around it, and an address because it looks like somewhere you could post a letter. On a line naming a person, a street, an email, a phone number, a card number, a date and an API key it boxed all seven. It is not a guarantee of anything: it also boxed a reference number that was nobody’s business but the company’s. Read every page, including the ones it marked nothing on.
- Why is the finding a separate button?
- Because it is a 875 MB download and about a gigabyte of memory while it runs, and nobody should pay that for opening a document. Marking pages by hand needs none of it and the blacking out is exactly as permanent either way. Press the button only when you want the reading done for you, and your browser keeps the download after the first time.
- Why does it say the finding needs a desktop?
- Because a phone will not give a browser tab a gigabyte. It would download the whole thing and then be killed, which is a worse experience than being told up front. The rest of the tool works on a phone exactly as it does anywhere else.
- Why does it black out more than the word I meant?
- Because it covers the whole run of text the match sits in. A PDF reports how wide a run is and not how wide each letter in it is, so trimming to the exact characters would be an estimate, and an estimate that comes up short leaves the last digit of a card number showing. Covering a few extra words is the safer mistake. Remove the box and draw a tighter one if you would rather.