Redact personal data
Find names, addresses, phone numbers, e-mail addresses, IBANs and other personal data in PDFs, scans, photos and text – and black them out with one click. Lift what should stay; mark what was missed yourself. The result is rebuilt as an image so nothing is left under the bars. Nothing is uploaded.
How it works
Document or text
Drop a PDF, scan, photo or screenshot into the box – or paste text in the “Text” tab.
Check
Found spots are blacked out at once. Tap a bar to lift it, tap a word to redact it, or drag a rectangle over an area.
Save
Save the redacted PDF or image. Every page is rebuilt as an image: there is no text left under the bars, and metadata such as the author is gone.
Frequently asked questions
Are my documents uploaded?+
No. Files and text are only read, searched and redacted in your browser. Text recognition and the PDF reader come from mytoolio.org itself. Only if you switch on AI detection is the recognition model loaded once from Hugging Face – your content is not sent anywhere.
What is found automatically?+
E-mail addresses, phone numbers, addresses (street with number, postcode with town), IBANs (with check digits), credit card numbers (with checksum), tax IDs, social security numbers, IP and web addresses, dates if you like – plus names after Mr/Mrs/Ms and values in form fields such as “Name:”, “Address:” or “Date of birth:”. AI detection adds names of people, places and companies anywhere in the text.
Is the redaction really safe?+
Yes, for the saved result: every page is rebuilt as an image with the black bars burnt in. Unlike a black rectangle that is only placed over the text, nothing underneath can be selected or copied. The new PDF carries no metadata of the original either. Still check every page – automatic detection can miss spots.
What does AI detection do?+
A small language model (DistilBERT, multilingual, AFL-3.0 licence) recognises names of people, places and companies without a fixed pattern – in German, English and eight more languages. It is loaded once when you first switch it on (about 135 MB) and then runs offline in your browser.
Can I redact text without a file?+
Yes, in the “Text” tab: paste text, check the hits and have them replaced by bars (█████), placeholders like [NAME] or ***. Tap hits that should stay.
Is the PDF still searchable?+
No. Because every page is rebuilt as an image, the redacted PDF contains no selectable text – which is exactly what makes it safe. If you need searchable text again, run the result through the text recognition.
Does this make a document GDPR-compliant?+
The tool helps with anonymising because the data never leaves your device. Whether a document is truly anonymous afterwards depends on its content – context such as job titles or rare details can still identify a person. So check the result yourself.
Text recognition with Tesseract.js (Apache 2.0), PDF handling with pdf.js and pdf-lib (Apache 2.0 / MIT), name recognition with Davlan/distilbert-base-multilingual-cased-ner-hrl (AFL-3.0). Automatic detection does not find every spot – please check before sharing.