Find and mask personal data in a dataset.
How to use PII Scrubber
Add a dataset. Drop a .csv or .jsonl file onto the uploader or pick it from your device. Only one file is accepted at a time.
Choose a masking style. Select Partial to keep a hint such as j***@example.com, or Full redaction to replace the whole value with a labelled placeholder.
Run the scrub. Press Scrub PII. The tool scans every cell or string value and masks matches.
Review what was found. The What was found panel shows a count for each type of value that was masked.
Download the result. Press Download scrubbed file to save a copy named after your file with -scrubbed added.
Limits and privacy
- The maximum file size is 25 MB, and only .csv and .jsonl files are supported. Parquet and Arrow are not.
- Detection is pattern-based and runs in your browser. It looks for email addresses, US Social Security numbers (###-##-####), credit card numbers, phone numbers and IPv4 addresses.
- Card numbers must be 13 to 19 digits and pass a Luhn checksum to be masked. Phone patterns are US-style, such as 415-555-0199 or (415) 555-0199.
- Names, street addresses, dates of birth, account numbers, national IDs from other countries and free-text identifiers are not detected.
- It cannot guarantee that all personal data is found. Review the output before sharing it.
- For JSONL, string values inside valid JSON lines are scanned and the line is rewritten as compact JSON. Object keys are not scanned. A line that is not valid JSON is scanned as plain text.
- For CSV, the file is re-written with CRLF line endings and quoting only where needed.
- Without an account you get 3 free actions on this tool every 24 hours; signing in removes the cap.
Frequently asked questions
What kinds of PII does it detect?
Email addresses, US Social Security numbers, Luhn-valid credit card numbers, US-style phone numbers and IPv4 addresses. Nothing else is detected.
Will it catch every piece of personal data?
No. It matches fixed patterns, so names, addresses and many other identifiers are left as they are. Treat it as a first pass and check the output.
Is my data uploaded?
No. The file is read and scrubbed locally in your browser, and the download is generated there.
What is the difference between partial and full masking?
Partial keeps a small hint, such as the first letter and domain of an email or the last four digits of a card. Full redaction replaces the value with a placeholder like [EMAIL_ADDRESS_REDACTED].