Redact a PDF

Because a PDF keeps its text separately from what is drawn on the page. A rectangle is another object placed on top; the words underneath are untouched and any reader, or the PDF-to-text tool on this site, will return them. We measured it while building this: a document covered with a black rectangle took 46 milliseconds to produce and handed back the hidden line verbatim. Redactions of exactly this kind have been undone in published court filings more than once.

Each page you mark is rendered to an image, the black boxes are painted onto that image, and the page is replaced with the picture. There is no text layer left on those pages, so there is nothing to extract. Pages you did not mark are copied across untouched and keep their text — rasterising the whole document would work too, but it would throw away searchable text you never asked to lose.

That page can no longer be searched or selected, its links and form fields are gone, and the file grows — in our measurement a text-heavy document went from 7 KB to 170 KB with one page rasterised, against 971 KB if every page had been. That is the trade: a page that is genuinely clean is a page that has become a picture. Any tool offering redaction without that cost is covering rather than removing.

Type the phrases that must not survive and, once the new file is built, its text is extracted and searched for them. You get a line per phrase saying GONE or STILL THERE. It is a small thing and it is the whole point: a claim that something was redacted is worth much less than a search of the finished file showing it is not in there. If you name nothing, you are told that nothing was checked rather than being given a green tick.

No. Everything — rendering, covering, rebuilding and the check — happens in this page on your own machine. A document you are redacting is by definition one you do not want other people to have, so sending it to a server to be made private would be a strange way round.

Because a PDF keeps its text separately from what is drawn on the page. A rectangle is another object placed on top; the words underneath are untouched and any reader, or the PDF-to-text tool on this site, will return them. We measured it while building this: a document covered with a black rectangle took 46 milliseconds to produce and handed back the hidden line verbatim. Redactions of exactly this kind have been undone in published court filings more than once.

Each page you mark is rendered to an image, the black boxes are painted onto that image, and the page is replaced with the picture. There is no text layer left on those pages, so there is nothing to extract. Pages you did not mark are copied across untouched and keep their text — rasterising the whole document would work too, but it would throw away searchable text you never asked to lose.

That page can no longer be searched or selected, its links and form fields are gone, and the file grows — in our measurement a text-heavy document went from 7 KB to 170 KB with one page rasterised, against 971 KB if every page had been. That is the trade: a page that is genuinely clean is a page that has become a picture. Any tool offering redaction without that cost is covering rather than removing.

Type the phrases that must not survive and, once the new file is built, its text is extracted and searched for them. You get a line per phrase saying GONE or STILL THERE. It is a small thing and it is the whole point: a claim that something was redacted is worth much less than a search of the finished file showing it is not in there. If you name nothing, you are told that nothing was checked rather than being given a green tick.