Choose a file to begin.
Sensitive Info Remover: before and after
An illustration with sample files; your own files show their own results.
BeforeSensitive data, easy to read and copy
Card number and IBAN
SSN, phone and email
Passport MRZ lines
AfterDestroyed for good, not just covered
Pixels overwritten
No text behind the bar
Entropy proof included
Sensitive Info Remover — Permanently Remove Sensitive Data In-Browser
Finds and permanently removes sensitive data (card numbers, IBANs, bank account numbers, emails, phone numbers and more) in PDFs and images, entirely in your browser. Nothing is uploaded. It does not hide data behind a black box: it overwrites the raw pixels, so there is nothing left to peel back.
How redaction works here
- Detection (PDF): the text layer is extracted with pdf.js and checked by deterministic detectors: a real issuer prefix plus Luhn for card numbers, ISO 7064 mod-97 for IBANs, the RIB key for Moroccan and French account numbers, real phone-number shapes, range checks for IPv4. Table columns stay separate, so a date next to an amount is never read as one long number, and each hit maps to its exact glyph rectangle.
- Custom words: names and addresses have no fixed pattern, so you type them and every occurrence in the document is redacted.
- Destruction: matched regions are painted into the pixel buffer with solid black or cryptographic noise from
crypto.getRandomValues(). The original values are gone. - Rebuild (PDF): each page is flattened and re-assembled into a fresh PDF with no text layer, so a “black box over selectable text” leak is impossible. The output is image-only and not searchable.
- Metadata: a clean canvas and new document drop the source EXIF, XMP, GPS and PDF Info.
- Proof: the tool measures the entropy of each redacted region and re-extracts text from the output to confirm nothing is left.
Limits
There is no bundled OCR or vision model, so automatic detection works on PDFs with a text layer, not on photos or scans; for those you draw the boxes and the destruction steps are identical. Detectors catch well-formed, checksummed identifiers and miss data written in prose, unusual national IDs, or anything inside an image; add names and other free text as custom words. Review the highlighted regions before exporting: you decide what is sensitive.
Frequently asked questions
Why do black boxes drawn in a PDF reader not count as redaction?
A drawn box is only an overlay. The original text stays in the PDF underneath and can be copied or extracted. This tool deletes the glyphs and rebuilds the page, so the text is no longer in the file.
What sensitive data does it detect on its own?
In a PDF's text layer: card numbers (issuer prefix + Luhn), IBANs (mod-97), Moroccan and French RIB account numbers (key-checked), long ID and reference numbers, US SSNs and EINs, emails, phone numbers, IPv4 addresses and passport MRZ lines, plus any name or word you type. You review every hit before anything is removed.
How do I redact a bank statement before sending it to a landlord or lender?
Open the statement PDF and the tool flags card numbers, IBANs, RIB and long account numbers in its text layer. Type any name or address you also want gone, review the hits, and download a rebuilt PDF with no text left under the boxes. Check with whoever asked for the statement which details they need to see.
Can the redacted text be recovered afterwards?
No. The page is flattened and rebuilt without the removed glyphs, so no text layer survives. In images, the covered pixels are overwritten with solid fill or cryptographic noise and the metadata is stripped.
Does it work on scanned documents?
Yes, manually. No OCR model is bundled, so you draw the regions and the pixels underneath are destroyed. That means nothing is uploaded and you get no false sense of automatic coverage.
How do I prove to someone that a region was really destroyed?
The tool measures the entropy of each redacted region after the overwrite. Noise-filled regions read near maximum entropy, which is evidence the original content is not recoverable from those bytes.
Is my file uploaded anywhere?
No. Parsing, detection, pixel destruction and export all run in your browser. No server, model host or analytics endpoint is contacted.
Why is the redacted PDF no longer selectable or searchable?
Because every page is flattened to an image and the original text and fonts are discarded, which is what makes the redaction irreversible: text that could be selected could still be leaked. Use a higher DPI for a crisper result.
What is the difference between solid black and noise fill?
Both destroy the underlying pixels. Solid black gives a uniform region; cryptographic noise fills it with high-entropy random bytes, which shows the area was destroyed on purpose and hides the original shape.
Sensitive Info Remover at a glance
| Where it runs | Entirely in your browser. Files are never uploaded. |
|---|---|
| Cost | 90 tokens per file (about $0.90 at the base rate) |
| Accepts | PDF, PNG, JPEG, WebP |
| Produces | A rebuilt PDF with no text layer behind the redactions, or an image with the pixels overwritten, plus an entropy report for each destroyed region. |
| Batch | One at a time. |
| Account needed | Only to spend tokens. Browsing and previewing are free. |
| What it does not do | Automatic detection needs a real text layer. Scans and images have no OCR here, so their regions are marked by hand. |