Redact PDF

Visual cover draws an opaque black rectangle over the area as an annotation layer — the underlying text and image data are still inside the PDF, just hidden visually. Anyone can flatten the file, delete the annotation, or run pdftotext on it and recover the "redacted" content. True redact rasterises the entire affected page to a high-DPI image with the black box burned into pixels, then replaces the original page with that image. The original text, fonts, and vector content beneath the box are physically removed from the document. Use true redact for anything you genuinely need to hide — names, account numbers, addresses, signatures, IDs. Visual cover is only acceptable when you want a "preview" mark that the recipient can remove themselves.

With true redact, no — the page becomes a flat image after redaction, so the original text characters, embedded fonts, and metadata behind the black box no longer exist in the file. Even forensic tools cannot reverse a rasterised page; you would need the original unredacted PDF. With visual cover the redaction is fully reversible: anyone with Acrobat, a free editor, or a copy-paste into a text tool can lift the box. This is why we default the mode to true redact and place a clear warning beside the cover option. We strongly recommend never sharing a cover-only file with anyone outside your machine.

Yes. Pick the Auto-find emails/SSN/phone mode and click Scan & mark — the tool reads the text layer of every page through PDF.js, runs three regex patterns (email RFC 5322, US SSN ###-##-####, North-American phone with optional country code), and adds a redaction box around each match. You can review every detected area on screen, delete false positives with the ✕ badge, then run true redact to burn them out. Auto-detection only works on PDFs with a real text layer; for scanned image PDFs run OCR first using our OCR PDF tool, then come back and redact.

No. The PDF you pick is read into your browser via FileReader, opened locally with pdf-lib and PDF.js (both WebAssembly), and the redaction — including the high-DPI page re-rasterisation for true redact — runs on a canvas inside your tab. The result is saved to a Blob and offered as a download without any network call carrying your document. You can verify in DevTools Network: no outbound POST with your PDF. This is the only correct architecture for redaction; uploading a file you want to redact to a third-party server defeats the purpose entirely.

Inside the redacted (rasterised) pages, yes — page-level metadata such as comments, form fields, hidden layers, and tagged-content beneath those pages becomes part of the flat image and is destroyed with the rest. Document-level metadata (author, title, creation date, XMP) is preserved by default; if you want a clean export, run the result through Flatten PDF afterwards to strip annotations everywhere or use a metadata-clean tool. Bookmarks targeting redacted pages keep their text label but point to the new image page. If you need a true "scrub everything" workflow, redact first and flatten second.

Affected pages are re-rasterised at 1400 px wide, which translates to roughly 170–180 DPI on a Letter/A4 page. That is dense enough for the body text and most figures to look indistinguishable from the original on screen and acceptable on a normal home printer. For archival or print-shop output you may notice a slight softening on the redacted pages compared with un-redacted vector pages. Pages with no redaction boxes are not touched at all — they stay as the original vector content, perfectly sharp. If you need higher DPI for print, redact, then re-rasterise the whole document with a dedicated PDF-to-image tool.