Redact PDF
Delete the words, not just the view of them. Your files never leave your device.
Drop a PDF here, or click to browse
PDF — the marked text is deleted, not covered over
A black box is not redaction
Draw a filled rectangle over a name in almost any PDF editor and it looks redacted. It is not. The rectangle is a shape drawn on top; the text underneath is untouched, and selecting the area, copying it, or running any text extractor reads it back out in full.
This is not a theoretical risk. It is how names have been recovered from court filings, how figures have been read out of published settlements, and how supposedly anonymous documents have been de-anonymised — in each case by people who simply pressed Ctrl + A and Ctrl + C.
What this does instead
The page's content stream — the actual list of drawing instructions that makes up the page — is edited. The glyphs inside each marked area are deleted, and the surrounding text is left exactly where it was, so nothing shifts and nothing slides out from under the box. Then an opaque rectangle is painted into the page, not added as an annotation, because an annotation can be deleted in any reader.
Scanned pages get different treatment for the same reason: when a marked area overlaps an image, the page is flattened, because a box over a photograph leaves the original pixels sitting in the file.
And then it checks its own work
Removal is attempted, and then proven. The finished PDF is re-opened, its text is extracted again with a different engine from the one that wrote it, and every marked rectangle is checked for anything still readable inside it.
A page that passes keeps its selectable text. A page that fails is rasterised — which cannot leak text even in principle — and you are told exactly which pages that happened to, and why. The result is that leaving something behind is not a way this tool can fail quietly. The worst case is a page that loses its selectable text and says so.
Before and after redacting
Redaction removes content, so the file usually gets a little smaller — but if you also need it under an upload limit, Compress PDF shrinks images without rasterising your text. If you only need to share part of a document, it is often safer to extract those pages first with Organize PDF and redact what remains. And if the document started life in Word, Word to PDF converts it here rather than on someone else's server.
Redaction does not touch document metadata such as the author name or the software that produced it. To strip that as well, run Remove Metadata afterwards.
Related tools
Frequently asked questions
- Is a black rectangle over text enough?
- No, and this is the single most important thing to understand about PDF redaction. Drawing a black box in most editors adds a shape on top of the page while leaving the text exactly where it was. Anyone can select it, copy it, or run a text extractor and read it straight out. That mistake has exposed names in court filings and figures in published reports. This tool deletes the text from the page's content stream, so there is nothing left underneath to recover.
- How do I know it actually worked?
- Because it is checked, not assumed. After the marked content is removed, the finished PDF is re-opened and its text re-extracted, then compared against the areas you marked. If anything is still readable inside one of them, that page is flattened to an image instead — which cannot leak text even in principle — and you are told which pages that happened to. Under-redaction is not a way this can fail silently.
- Will the rest of my document stay searchable?
- Yes. Only the marked areas are touched; text everywhere else stays selectable, searchable and accessible. That is deliberate — a redaction tool that flattens the whole document to images to solve one rectangle has destroyed the file to protect one name.
- Is my file uploaded?
- No. Your files never leave your device The document is opened, edited and rebuilt entirely in your browser. Given that people redact contracts, medical records, bank statements and legal filings, sending the unredacted original to someone else's server to have it redacted would rather defeat the point.
- Does it handle scanned documents?
- Yes, and carefully. A scan is one large image per page, so a black box over it would leave the original pixels in the file for anyone to extract. When a marked area overlaps an image, that page is flattened so the underlying pixels are genuinely gone. Scans also often carry an invisible OCR text layer — fully copyable but not visible — which is removed along with everything else.
- What about comments, form fields and links?
- Annotations that overlap a marked area are removed too. A sticky note, a filled form field or a link carries its own text quite separately from the page, so painting over the page would do nothing to it. Annotations elsewhere on the page are left alone.
- Can I undo a box before redacting?
- Yes. Click any box to remove it, or clear them all. Nothing is changed in the file until you press the redact button — and the redaction applies to a copy you download, never to your original.
- Is this as good as Adobe Acrobat's redaction?
- It removes content the same way — by editing the page's content stream rather than covering it — and it adds a verification pass that Acrobat does not: re-reading the finished file to prove the text is gone. What it does not have is Acrobat's search-and-redact-all-instances workflow, so on a long document you mark the areas yourself.