Redact a PDF — remove the text, don’t cover it

Why a black rectangle is not redaction

A PDF page is a list of drawing instructions: put these characters here, draw this line there. Drawing a black rectangle adds one more instruction on top — the characters underneath are still in the file, still selectable, still one copy-paste away from anyone you send it to. You can demonstrate this in ten seconds with any PDF whose text was "hidden" that way: select across the black bar, copy, and paste into a text editor. The text falls out, byte-identical.

What this tool does instead

On pages with real text, marked words are removed by content-stream surgery: the drawing instructions for those exact glyphs are deleted from the page's instruction list. The secret is not obscured — it is gone, and every other word on the page remains real, selectable text. Deleting text cannot destroy pixels, so a picture under a mark, a scan or a photo, has its marked pixels overwritten inside the picture itself; the rest of the picture is kept as stored. A drawing under a mark, a line, a box or a chart, is replaced by a picture of itself with the marked part removed; everything else on the page stays as it was. Where the tool cannot do that and check it (a drawing nested inside another drawing, or a picture the tool cannot read, for example), it rebuilds the page as an image instead. That path has a stated cost: the page won’t zoom as crisply, and the file grows. Its remaining text stays selectable and searchable, and its links and notes keep working. Untouched pages keep everything, and the receipt tells you exactly which pages took which path.

Verified twice — and one thing only you can verify

After redacting, the tool extracts the text back out of its own output and checks the marked content is not in it; then it renders the output and checks the marked region is opaque. On a page where pictures or drawings were cut, it also reads each of them back to check nothing is left under a mark, and checks the rest of the page looks as it did. If any check cannot pass, the result is thrown away and you are told — you never get a page that claims to be redacted and is not. Files attached to the PDF are not part of any page and are carried over as they were: Verify lists them, and Sanitize removes them along with the rest of the hidden data. One check remains yours: the tool can prove the mark is opaque, but only you can confirm it covered the right thing. That is why your marks snap to the words they cover, why the mark list names what each mark removes, and why the preview shows the exact region that will be destroyed before you commit.

Scans that were OCR'd

A scan that went through text recognition carries the recognized text as an invisible layer over the picture — the secret exists twice. Deleting only the text layer would leave it plainly readable in the image, so on such pages the marked pixels are always destroyed as well. This is enforced in the engine; there is no way to switch it off, on purpose.

Questions people actually ask

Why does a redacted PDF sometimes still show the text?

Because covering is not removing. A rectangle drawn over text is just one more drawing instruction; the characters are still in the file underneath it and can be selected, copied, or extracted. Redaction has to delete the text from the file itself, which is what this tool does — and then proves by extracting its own output.

Can text be recovered from a PDF redacted here?

Run the check yourself — the page offers to extract the text of your redacted file right after committing. Marked content comes back as zero characters. That check is also run automatically before the file is handed to you, and if it cannot pass, you get an error instead of the file.

What happens on a scanned page?

A scan has no text to delete, so the marked pixels are overwritten inside the scan itself, and the rest of the scan is kept as stored, apart from a thin band around each mark in a JPEG scan. If a scan is stored in a way the tool cannot change and check in place, the page is rebuilt as an image instead: it won’t zoom as crisply, and the file grows. Any recognized text on it stays searchable, except what you marked. Pages you didn’t mark keep everything.

Does the rest of my document stay selectable text?

Yes — only the marked glyphs are deleted; everything else remains real text, including on a page rebuilt as an image. Try selecting around the mark in the output.

Does redaction look inside files attached to the PDF?

No. Attached files are carried over exactly as they were. Verify shows what is attached, and Sanitize removes attachments along with the rest of the hidden data; run it after redacting, before you send the document.

Is my document uploaded for redaction?

No. Marking, removal, and both verifications run in your browser. The file never leaves your device — which for a document worth redacting is rather the point.