Redact a PDF so the words are gone.

A black rectangle is not redaction

The way most people black out a PDF is to draw a filled rectangle over the words. The page looks right. The words are still there, underneath, as selectable text — copy the region, or open the file in anything that reads the content stream, and they come straight back. Every so often this makes the news, usually with a court filing or a contract attached.

Redaction here works on the page's content stream instead. The instructions that paint those characters are found and removed, so there is nothing left to select, search or copy. The box you see afterwards is drawn as well, but it is a second line of defence for the reader's benefit — not the mechanism. If the box were deleted, the words would still be gone.

Text nested inside a form XObject — a reusable block a PDF can stamp in several places — is followed and rewritten too, down through eight levels of nesting. For an image, the covered pixels are decoded and overwritten rather than masked.

The saved file is rewritten, not appended to

This is the part that catches people out, and it is worth a paragraph. A PDF is normally saved incrementally: the new version is appended to the end and the old bytes stay in the file, which is how signatures survive an edit. It also means that if you redact and save that way, the removed text is still sitting in the previous revision, a few bytes back.

So redaction is the one edit that forces a full rewrite. Every other change in this app takes the incremental path; this one does not, because an edit whose entire point is that something becomes unrecoverable cannot keep a copy of it. There is a test that asserts a redacted page yields zero extractable characters afterwards.

How to redact a PDF

Open the document — in a browser tab, in the desktop app, or through the Chrome extension from wherever the file already is. Pick the redact tool, drag over what should go, and save. There is no sign-in, no upload step and no page limit, and redaction is free in all three.

The bar left behind is white by default, which reads as a gap in the page rather than as the familiar black bar. Worth knowing before you send the file on, in case the person receiving it expects to see something struck out.

What it does not clean

Redaction here rewrites what is drawn on the page. It does not touch the document's properties, its XMP metadata, attached files, embedded JavaScript, or hidden optional-content layers. A document can carry an author name, an original filename or a layer that is switched off but still present, and redacting the visible text leaves all of that where it was.

For most documents that is fine. For a file going to an opposing party or into a public record, check those separately. We would rather say this plainly than let you find out from someone else.

Why not an online redaction tool

The usual answer to “how do I redact a PDF” is a website that asks you to upload it first. Consider what that means: to have the sensitive part of a document removed, you send the document, sensitive part included, to somebody else's server. The secret you are trying to contain has already left the building.

Nothing here is uploaded, because there is nowhere for it to go. The web app does the work in the page itself, and the Chrome extension declares no permissions and no host permissions at all — the browser will not let it reach a server even if something in it tried. The source is public, and the extension's manifest is short enough to read in a minute.

Questions