Nothing is uploaded. When you drop a PDF, the browser File API reads it into memory, pdf.js renders each page to a canvas so you can see it, and when you apply the redactions pdf-lib rebuilds the document on your own machine and hands it back from a local Blob for download. No bytes are sent anywhere. That matters because the documents people redact — contracts, bank statements, medical records, court filings, ID scans — are exactly the files you would never want to hand to a cloud service. There is no login, tracking or CDN in the processing path, the tool works offline once cached, and closing the tab discards everything.

This is true redaction, not a cover-up. Many “black box” edits in PDF viewers add an annotation that floats over the text — the recipient can move or delete the box, or copy the text out from underneath. Here, when you press Redact permanently, every page is re-rendered to a high-resolution raster image with the redaction rectangles painted onto the pixels, and pdf-lib assembles a brand-new PDF from those images. The original selectable text layer is discarded in the process, so anything beneath a bar is gone for good: it will not appear in copy-paste, text search, or extraction tools. The trade-off is that the output pages are images, so the whole document is no longer selectable text — that is the price of guaranteed removal.

The tool reads the PDF’s text layer with pdf.js and offers two assisted modes. Find & Redact searches for a specific word or phrase and reports how many matches it found across how many pages. Auto-Detect PII runs a set of pattern matchers over the text — email addresses, phone numbers, credit-card numbers, Indian Aadhaar and PAN numbers, dates and currency amounts — and lists each category with a count so you can bar them in one tap. These modes place a redaction bar across the region of each page that contains a match, which you should visually check and nudge before applying, since they cover the matching line rather than pinpoint pixel-perfect word boxes. For exact control you can always draw boxes by hand. Scanned image-only PDFs have no text layer, so auto-detect and search cannot read them — use manual boxes there.

PDF Redact is free with no sign-up, watermark or page limit, and because it is 100% client-side it works in any modern browser on Android, iPhone, iPad, Windows, macOS or Linux with nothing to install. The page viewer supports touch, so you can drag redaction boxes with a finger, switch pages with the thumbnail strip, and see a per-page count of how many areas are marked. You pick the bar colour (black, white, gray or red), undo the last box, remove an individual box or clear them all before committing, and only then are the redactions burned in. One PDF is processed at a time and the practical ceiling is your device’s memory, so a very large or high-page-count file is bounded by available RAM.