pdftoolskit.org
PDF (Portable Document Format) utilities, in the browser
Say hi →

Redact PDF — and Verify the Text Is Actually Gone

drag or scan for patterns · text under the box is deleted · verified after saving

Drop a PDF and either drag rectangles by hand or have the tool find sensitive patterns — emails, phone numbers, national IDs, credit cards, IBANs, or any regex — and box every match across every page. Then the part that matters: on the pages you redacted, the text underneath is deleted, not covered. Afterwards the saved file is read back with a separate PDF engine and the tool tells you whether any redacted text can still be extracted.

This is the difference between a redaction and a rectangle. A black box drawn over a name leaves the name in the file, selectable and copy-pasteable by anyone who drags across it — which is how court filings, government releases and corporate disclosures keep leaking the exact thing they meant to hide.

Drop your PDF file here

One PDF at a time · drag rectangles, or scan for patterns below

    Redact by pattern — scans every page and adds a black box over each match
    No scan yet.
    —
    no files
    Ready.

    When to use this tool

    Before sharing a contract that carries another party's pricing, a bank statement going to a landlord, a medical record going to an insurer, a document produced in response to a records request, or a screenshot with a customer's email in it. The common thread is that the file has to go out but part of it must not.

    Why a black rectangle is not a redaction

    A PDF page is a list of drawing instructions. "Show the text Jane Doe at this position" and "fill this rectangle with black" are two separate instructions, and drawing the second does nothing to the first — it just happens to be painted later, and therefore on top. The name is still in the file. Select across the box and paste, or run any text-extraction tool over the page, and out it comes.

    This is not a theoretical risk. It is the single most repeated document-handling failure there is: redacted court exhibits, sealed filings, government disclosures and corporate submissions have all been un-redacted by readers who simply pressed Ctrl+A and Ctrl+C. Any tool that draws a box and calls the job done is setting you up for it.

    What this tool does instead

    In the default mode, every page you drew a box on is rendered to a bitmap with the boxes painted in, and that bitmap replaces the page. There is no text object left underneath, because there is no text layer on that page at all — nothing to select, copy, extract or recover with a forensic tool.

    Only the redacted pages are converted. The rest of the document is copied untouched, so a 90-page report with two redacted pages keeps 88 pages of selectable, searchable text. That is the trade-off most tools get wrong in the other direction, by flattening the entire file and destroying its text layer to hide one phone number.

    Then it checks its own work. The saved bytes are re-opened with pdf.js and the text of every page is extracted. The tool reports whether any text survives on the redacted pages, and — if you used the pattern scanner — whether any of the specific values it matched can still be found anywhere in the output. You get a ✓ or a warning, not a promise.

    Redacting by pattern

    Hand-drawing boxes does not scale to a 200-page disclosure. The scanner takes the built-in patterns — emails, phone numbers, US SSNs, credit-card numbers, dates, IBANs, URLs, IPv4 addresses — or any regular expression you type, runs them over the extracted text of every page, and places a box on every match at the exact position of the text.

    Two things to know. The scanner deliberately over-redacts: when the PDF stores a matched string as several fragments, every fragment containing part of the match is boxed, so you may lose a character or two either side. And a pattern is not a guarantee of completeness — a phone number written "five five five" is not matched by any phone regex, so review the pages before saving. The boxes are editable: add, undo and clear as you go.

    How to redact a PDF

    1. Drop one PDF. The first page renders with a transparent overlay you can draw on.
    2. Pick the patterns you care about and click Scan PDF & add boxes to box every match across every page at once. Skip this if you are redacting by hand.
    3. Drag rectangles over anything else that must go, using Page ← / → to move through the document. Boxes are remembered per page, and Undo last box takes back a mistake.
    4. Leave the mode on "delete the text underneath" — that is what makes it a redaction. Raise the DPI if the redacted pages will be printed.
    5. Click Redact & download and read the verification lines: they say whether any text still survives on the redacted pages, and whether any matched value can still be found in the file.
    6. Open the result and try to select the redacted area before you send it. Thirty seconds of paranoia, every time.

    What redaction still cannot fix

    FAQ

    Does this permanently remove the text, or just cover it?

    It removes it. In the default mode each redacted page is replaced by a bitmap of itself with the boxes painted in, so there is no text layer on that page and nothing to extract. The tool then re-reads the saved file and reports whether any text survived, so you are not taking that on trust.

    Will the whole document stop being searchable?

    No — only the pages you actually redacted. Pages with no boxes are copied as-is and keep their text, links and vectors. That is deliberate: flattening an entire report to hide one email address is a bad trade.

    How do I redact every email address or phone number at once?

    Tick the pattern, click Scan PDF & add boxes, and every match on every page is boxed in one pass. The status line reports how many matches were found and of which kind, so you can sanity-check the count before saving.

    Can I redact a scanned PDF?

    You can draw boxes on it and the result is genuinely redacted, since the page was already an image. Pattern scanning, though, needs a text layer — run OCR PDF in searchable-PDF mode first, then bring that file back here.

    What DPI should I pick for the redacted pages?

    200 is a good default and keeps text crisp on screen. Use 300 or 400 if the document will be printed or archived; the redacted pages get larger, but only those pages.

    When would I ever want the "black box only" mode?

    When the file must stay fully searchable and the box is cosmetic — hiding a spoiler, tidying a screenshot for a slide. Never for personal data, and the tool says so in red when you use it.

    What if a match spans several text fragments?

    Every fragment containing part of the match is boxed, so slightly more than the literal match may be covered. Over-redaction is the safer default, and you can see exactly what was boxed before you save.

    Are my files uploaded?

    Never. Rendering, redaction and verification all happen in your browser via pdf.js and pdf-lib. Given what people redact, a tool that uploaded the unredacted original first would be self-defeating. See the privacy policy.