Privacy ToolboxbyBoxfish Labs

Text redact

Clean text before you share it with AI or third parties.

0 / 8000 characters

Your text and results are available only while this page remains open. Refreshing, closing, or leaving the page clears the current session.

Identifier coverage

International protection is enabled by default. Add a region or country profile if the text may contain national IDs, tax numbers, local addresses, licences, vehicle registrations, or other local formats.

Country-specific profiles

Add tested rules for national identifiers, tax numbers, postcodes, licences, and local address formats. They supplement international protection; they do not replace it.

France (Coming soon)United Kingdom (Coming soon)United States (Coming soon)

What should be detected and handled?

Choose which kinds of information to detect. Recommended protection is a safe default; select all only when you need deeper context reduction.

Enables every category, including organisations, dates, and sensitive context. May reduce readability; you can restore individual findings before copying.
Turns on direct identifiers: names, contact details, addresses, government IDs, financial data, online identifiers, and secrets. A practical default for most text.
People’s names and personal references such as “my client Anna Schmidt”. Helps protect identity while keeping narrative structure with placeholders.
Email addresses, telephone numbers, and similar ways to reach a person directly.
Postal addresses, street names with numbers, postcodes with city names, and precise coordinates.
National ID numbers, tax identifiers, passports, driving licences, and similar official document numbers from enabled coverage profiles.
Payment-card numbers, IBANs, BICs, and related payment details.
IP addresses, URLs that may contain tokens, and social-media handles where personal.
API keys, passwords, JWTs, private keys, connection strings, and other credentials that must never be shared.
References to health, politics, religion, ethnicity, disability, union membership, or similar sensitive context that may need review.
Company, school, employer, clinic, and project names. Optional because removing every organisation can make text hard to use.
Dates of birth, distinctive dates, case or order numbers, and other contextual identifiers that can contribute to re-identification.

Automated detection may make mistakes. It can miss personal, sensitive, confidential, or identifying information, and it may incorrectly flag or remove content that is not personal data. The cleaned output is not guaranteed to be anonymous, free of personal data, legally compliant, or safe to share. Review all results carefully—including contextual details—before disclosing text to an AI system, third party, or public audience.

Paste text, choose what to protect, then detect.