PDF Personal Data Scanner

What it detects. IBANs (checked with the mod-97 checksum; a value that looks like an IBAN but fails is reported with low confidence), payment card numbers (Luhn checksum), BIC codes when a "BIC" or "SWIFT" label is next to them, email addresses, web addresses, phone numbers (international format with + or 00, or digits next to a label such as "Tel."), VAT identification numbers in the common patterns for Germany, Austria, France, the Netherlands, Belgium, Italy, Spain, Poland and Switzerland (pattern only, not looked up in VIES; the Polish NIP is also checked by its checksum), national identifiers with a published check digit (Polish PESEL, French NIR, Dutch BSN when labelled), identifiers next to labels such as "Steuer-ID" or "passport no.", dates next to birth-date labels, and capitalised words after a "Name:" style label.

Confidence. High means the value passes a published checksum. Medium means a strict format or a looser format next to a label. Low means a format only, a bare name label, or a checksum that fails. Names in running text are not detected, because no reliable pattern exists.

The report. Results show page number, category, the matched value, the context, the confidence and the reason. Values are hidden by default and shown only if you tick the box; the downloadable CSV is masked unless you choose otherwise, and spreadsheet-formula characters at the start of a cell are neutralised. A second download is a redaction list in JSON with page numbers and approximate areas, without the values, which you can use to find and remove the items.

A safe workflow. Scan, review each hit in the document, remove the data in the program that made it (or redact it so the text is truly deleted: a black box drawn over text usually leaves the text underneath), then scan the result again. Pages with no text layer, such as scans, cannot be scanned; the tool tells you how many there were.

PDF files up to 50 MB and 300 pages with a text layer. Encrypted PDFs are not opened. Scanned pages (images without text) are counted but not read; run text recognition first, for example with the PDF Text Editor, then export and scan the text version.

1. Drag in the PDF. It is read and scanned in your browser.
2. Read the list. Tick Show the matched values if you need to see them.
3. Check each finding on the page itself; ignore false alarms and note what is missing from the list.
4. Download the masked CSV report or the JSON redaction list if you want to work through the findings.
5. Remove the data, then scan again.

• Checking a document before you email it to a third party
• Reviewing a report exported from a system for private numbers that should not be in it
• Preparing a list of items to redact in a contract or invoice bundle
• Spotting a private IBAN or phone number in a public PDF you manage
• Teaching colleagues what personal data looks like inside a file

This finds patterns, not meaning. It will miss data written in unusual ways (for example an IBAN split across lines with extra characters, or digits spelled out), data inside images, and names in ordinary sentences. It will also flag harmless numbers that happen to look like an identifier, which is why each finding carries a confidence and a reason.

It is not a GDPR compliance tool or certification. Whether information is personal data, and whether you may share it, depends on the law and your situation. The tool does not redact anything; it only lists.