# Redact personal data

> Redact PDFs and images for real, or anonymize text before pasting it into ChatGPT: emails, phones, cards, IBANs and ID numbers become placeholders. No upload.

URL: https://nibbo.org/en/redact-pii/

## How it works

1. Detectors check checksums where the format has one: Luhn for cards, mod-97 for IBAN, Verhoeff for Aadhaar, and the official algorithms for CPF, CNPJ, INN, SNILS and TC Kimlik. That keeps false alarms low.
2. Each value gets a stable placeholder like [EMAIL_1]: the same address is always [EMAIL_1], so the AI can still follow who is who. The key to undo it stays in this tab.
3. PDF pages are redrawn as images with the black boxes burned in, so the hidden text is really gone, not just covered.
4. For screenshots you draw boxes or blur by hand. Optional OCR (Tesseract) finds text to hide automatically.

## Questions

### Does it catch everything?

No automatic tool does. It finds structured data well; names, addresses and free-form details are harder. Turn on “Names” or add your own words, and always read the result before sharing.

### Why do redacted PDF pages become images?

Drawing a black box over text in a PDF only hides it visually: the text underneath can still be copied or extracted. Rasterizing the page removes it for good. The trade-off is that the result is not searchable and is larger.

### Does OCR upload my screenshot?

No. The OCR engine and its language data are downloaded once from this site (about 5–15 MB) and kept for offline use; the recognition itself runs on your device. OCR is optional and only loads when you click it.

### Where is the key that restores my data?

Only in this browser tab. You can save it as a JSON file; if you close the tab without saving, the placeholders can’t be reversed.

## Why Nibbo

- Runs on your device, nothing is uploaded
- No sign-up, no watermark
- No queue and no daily limit
- Works offline once opened
- Free for files up to 3 GB
- [Nibbo vs Canva, iLovePDF, Smallpdf](https://nibbo.org/en/guides/nibbo-vs-canva/)
