Docuboxer

Redact PDF

Permanently black out names, IDs, account numbers or sensitive text in a PDF, verified in the final file.

100% local — your files are never uploaded to any serverNo signup · Free
Also in PDFSanitize PDFView & Edit PDF MetadataFlatten PDFCheck if a PDF Is DangerousUse it in the Suite with ⌘K and your history →

Drop your PDF here to permanently redact sensitive data

Or click to choose a PDF file from your device (max 100 MB, 200 pages)

Type and size are checked when you drop it

How do I permanently redact sensitive text and data in a PDF?

To permanently redact a PDF, drop your file here, mark the text or areas you want to remove, click “Redact PDF”, and download the file once the anti-leak verification report turns green. Drawing a black rectangle on top of text in macOS Preview, Word, or a comment tool does not redact anything: the original characters remain right underneath in the page stream, and anyone can copy-paste them, search them with Ctrl+F, or extract them in seconds. This tool runs PDFium compiled to WebAssembly inside your browser to strip the underlying text glyphs, vector paths, and image pixels within each marked box, paints the black redaction mark over the emptied region, and rebuilds a clean PDF document without metadata (/Info or XMP), annotations, form fields (/AcroForm), optional content layers (/OCProperties), structure trees (/StructTreeRoot), or previous incremental save history. Before enabling the download button, an independent verifier inspects the output PDF across several layers (extractable text via pdfjs, internal objects via pdf-inspector, decompressed streams, raw bytes in ASCII, UTF-16BE, and hex, and, for boxes drawn by hand, a position check that looks for extractable text under each box). If any trace of a redacted term remains, the download stays blocked. Everything runs locally on your device without uploading your PDF. Honest limits: no tool can claim “100% infallible” redaction on every malformed PDF, so instead we verify in your browser that the marked terms and the text under each box no longer exist in the downloaded file. The verifier does not read images: if the data is in a scan or a photo, check the preview before downloading, because there is no OCR to confirm it; also, on scanned full-page images, lossless re-encoding of modified bitmaps can increase file size (~×5 compared to a heavily compressed JPEG).

How to use Redact PDF

  1. Drop your PDF onto the upload box or choose it from your device (up to 100 MB and 200 pages). If the file is password-protected or encrypted, unlock it first in /unlock.
  2. Mark what you want to redact: draw rectangles directly on the page, search for literal text or regular expressions, or click the automatic pattern presets (SSN, credit cards with Luhn check, IBAN with MOD-97, Mexican RFC/CURP/CLABE/NSS, Spanish DNI/NIE, emails, and phone numbers).
  3. Review the pending redactions list by page. Click any item to jump to its page, undo the last mark, or remove individual boxes.
  4. Click “Redact PDF”. The PDFium engine removes the glyphs and pixels under each box, rebuilds a fresh document without metadata or incremental history, and runs the 5-layer verifier.
  5. Confirm that the verification report is green and check the preview, especially if the data is in an image, before clicking “Download redacted PDF”.

Privacy

Real privacy: Your files are never sent to a server, so no copy exists on any server, log or backup. Your browser keeps only the usage history (tool, date, file name and size, if any, and a generic error notice if something fails), and you can clear it in Recents.

Common use cases

Redacting contracts, court filings, or FOIA responses before publication

Infamous redaction failures happen when someone places a black box annotation over text without removing the underlying characters. Here the glyphs are stripped from the page content stream and verified across five inspection layers before you can download.

Removing SSNs, tax IDs, and national ID numbers from official records

Automatically spot and validate US Social Security Numbers (excluding invalid SSA ranges), Spanish DNI/NIE numbers (mod-23 check), and Mexican RFC, CURP, CLABE, and NSS identifiers in one click.

Blacking out bank account numbers, IBANs, and credit cards on statements

When sharing proof of address or payment receipts with landlords, accountants, or support teams, use built-in Luhn and ISO 7064 MOD-97 validators to locate and permanently erase card and account numbers.

De-identifying medical reports or HR files without cloud uploads

Strip patient names, phone numbers, email addresses, and case IDs locally in your browser so confidential health or personnel documents never pass through a third-party server.

Wiping hidden comments, form fields, and revision history alongside visual redactions

Rebuilding the output PDF into a brand-new document automatically drops sticky notes, links, AcroForm fields, author metadata, XMP packets, and older revisions stored in incremental PDF updates.

Frequently asked questions

Why isn't drawing a black box over text enough to redact a PDF?

PDF pages store drawing commands in layers. If you place a black rectangle over a sentence using a basic editor or annotation tool, the text operators stay in the content stream underneath the rectangle. Anyone opening the file can select the area, copy and paste the hidden words, or extract them with a script. True redaction deletes the underlying glyphs and image pixels from the PDF structure itself.

How does the 5-layer anti-leak verification work?

Before enabling the download, the verifier runs five anti-leak checks: pdfjs text extraction, pdf-inspector object scanning, annotation and widget inspection, decompressed FlateDecode stream scanning, and raw byte matching in ASCII, UTF-16BE, and hexadecimal. It also checks that /Info, /Metadata, /AcroForm, and /Annots are gone. A detected term in extractable text or a structural issue blocks the download. The verifier does not OCR text inside images.

Does it work on scanned PDFs and embedded images?

Yes. When a redaction box overlaps an embedded image or scanned page, PDFium changes its pixels in memory and re-encodes the image to destroy the covered area. On scans without a text layer, the tool cannot detect text inside the image automatically: draw a box manually and inspect the preview; the report does not replace OCR or visual review. “Flatten annotations and forms on all pages” incorporates those visible elements into each page's content, but does not rasterize the document or set a DPI resolution.

Why can a scanned PDF become larger after redaction?

Scanned PDFs usually store each page as a lossy JPEG stream. When a page image is modified to destroy the redacted pixels and saved back losslessly (FlateDecode), the file size can grow roughly fivefold (~×5). Whenever the tool detects full-page scanned images, it displays a notice comparing the original and final file sizes.

Are my PDF or search terms ever uploaded to a server?

Never. The page viewer, pattern detector, PDFium WebAssembly engine (served from the same origin), and 5-layer verifier all run inside your browser in a local Web Worker. Neither your PDF nor the text you search for ever leaves your device.

Can I redact a password-protected or encrypted PDF?

Not directly, because encrypted streams cannot be safely modified or verified without decrypting them first. If you know the password or the file only has owner permissions, remove the encryption first in /unlock and then open the unlocked PDF here.