Automatically redact personal data from a PDF
Scans a PDF's text for emails, phone numbers, card numbers and ID-style numbers, then shows every match for you to approve before anything is touched. Nothing is redacted automatically — you review the list, uncheck anything that isn't really personal data, and only the hits you keep get removed.
Auto-Redact PII
pattern detection · you approve every hit first
Loading tool…
How to Auto-Redact PII
- Add your PDF. Drag the file onto the drop zone. It's scanned for personal data immediately.
- Review each match. Every detected email, phone number, card number and ID number is listed with its category and page.
- Uncheck anything wrong. Pattern matching isn't perfect — deselect anything that isn't actually personal data.
- Redact and download. The pages with approved matches are rebuilt with that content genuinely removed.
How detection works
This runs the same text extraction Extract Text uses to pull positioned lines off each page, then checks each line against a set of patterns: email addresses, phone-number-shaped digit groups, card numbers that pass the same checksum card issuers use to validate them, and common ID formats like SSN-style numbers. Nothing here uses a database of real people or looks anything up — it only recognises shapes in the text.
That's a meaningful limitation, stated plainly: pattern matching will miss anything unusual — a phone number in an uncommon format, a national ID scheme it doesn't recognise — and can occasionally flag something that merely looks like a match, like a long reference number. That's exactly why nothing is redacted until you've reviewed the list.
The redaction itself is real
Once you approve a set of matches, this tool uses the same rebuild-as-image technique as Redact PDF: the page is rendered to a bitmap, the matched areas are painted out in the pixels, and the document is rebuilt from those images. The content is genuinely gone, not covered — with the same trade-off as manual redaction, that pages with any approved redaction lose their selectable text layer.
What this doesn't catch
Detection only sees text this site can already extract. Personal data inside a scanned image with no text layer, or written in a way that doesn't match any of the patterns above, will not be found. For a scanned document, run OCR — Searchable PDF first so there's text to scan, or fall back to Redact PDF and mark it by hand.
Frequently asked questions
Does this redact automatically, with no review?
No — every match is shown to you first, checked by default, and you can uncheck anything before clicking redact. Nothing is removed until you approve it.
What counts as an ID number?
Common patterns like SSN-formatted numbers (123-45-6789) and passport-style letter-plus-digit codes. It's a heuristic, not a database of real ID formats worldwide, so review these matches carefully.
Will it catch every card number?
It looks for 13-19 digit sequences that pass the same checksum real card numbers use, which filters out most random long numbers but can still miss an unusually formatted one or flag something that happens to pass the checksum by coincidence.
Are my files uploaded anywhere?
No. Scanning, review and redaction all happen entirely in your browser.
Does it work on scanned PDFs?
Only for whatever text layer already exists. A scan with no underlying text has nothing for the pattern matching to search — run OCR — Searchable PDF on it first.
How is this different from Redact PDF?
Redact PDF is manual — you draw every box yourself. This tool finds likely personal data automatically and lets you approve or reject each one, using the same underlying redaction technique either way.