Check a PDF before you send it

What a failed redaction looks like

It looks perfect. That is the whole trouble with it. A black rectangle sits over a name or a number, the page prints with the bar in place, and every person who opens the file sees exactly what the sender intended. Then somebody drags across the bar, copies, pastes into an email, and the name is there in plain type.

It happens because most PDF software has two separate features that produce the same black bar. One is a drawing or comment tool, which adds a shape on top of the page and leaves everything under it untouched. The other is a redaction tool, which takes the words out. Unless you know which one you used, the finished file cannot tell you by looking at it. Court filings and government releases have gone out with the first kind when the second was meant, which is why tools that hunt for exactly this exist.

How this check finds it

The checker does what your eye cannot. It reads where every run of text in the file sits, then draws each page and looks at the pixels in exactly those places. Text that can be read has ink and background in its patch of the page. Text that has been covered sits in a patch that is one flat color, black or gray or whatever the box was filled with. Text drawn in white on a white page is the same, just pale. Either way the words are in the file while the page says they are not, and that mismatch is what gets reported.

The report says which pages, roughly how many characters, and whether any of it is shaped like a Social Security, card or account number. It does not put the hidden words on the screen. If the file is yours, you already know what is under the bar and you need to know that it leaks; if it is not yours, reading it is not something this site helps with. The fix is one button: it opens the file in the redaction editor with bars already over every hidden patch, and the export rebuilds those pages so the words are gone.

What else a PDF carries

A redaction can hold and the file can still give things away. The metadata records name the author, the software and the dates, and sometimes an internal department or a username. Attached files ride along out of sight in most viewers. Scripts and automatic actions can run code when the file opens, launch a program, or send form contents to a server. Hidden layers carry content that viewers do not show by default. And a file saved with changes appended keeps its earlier versions inside it, where text that was edited out later can still be found.

Each of those is listed with what it is and where it came from. The cleaned copy the checker offers removes the metadata, attachments and scripts and writes the file as a single fresh version, without turning anything into an image, so the text stays selectable. It is then read back and checked the same way the original was.

Checking a file somebody sent you

If you receive a PDF and the checker finds text under its black boxes, the useful thing to do is tell the person who sent it, privately, before the file goes anywhere else. They may not know, and they may be able to recall or replace it. Forwarding it on, or reading what is hidden, turns their mistake into somebody else's harm. The check runs entirely in your browser, so the document itself goes nowhere while you look.

What it cannot see

A page that is nothing but a scan or a photograph has no text for the check to compare against the pixels, so it cannot tell you anything about that page beyond the fact that it is an image; look at those yourself. Text hidden behind a photograph rather than a solid shape is not flagged, because a photograph is not one flat color. And the check reads the file, not the meaning of the words, so it will not tell you whether a sentence that is plainly visible should have been removed. That part is still yours.

Common questions