PII Mask & Restore Tool De-identify before AI, restore after with one click
Step 1: Paste the original text containing personal data
Tip: Running auto-detect again resets the current masked result and mapping table. Do manual marking only after detection is finished.
What is the PII Mask & Restore Tool?
More and more people paste work content into ChatGPT, Gemini or Claude to get help, but that content often contains real personal data such as a customer's email, phone number, ID number or credit card number — pasting it means handing that data to a third-party server. This tool automatically detects common sensitive-data patterns in your browser, replaces each one with a code like [MASK_1] and [MASK_2], and generates an "original value ↔ code" mapping table so you can safely paste the masked text into any AI tool. Once the AI replies (the codes are usually quoted back as-is), you paste that reply here and use the same mapping table to restore the codes back to real data in one click. That's the core value of this tool: an AI chatbot can mask text for you, but without the mapping table it structurally cannot restore it — that's something AI alone can never do.
Supported detection formats
- Email addresses: standard email format.
- Phone numbers: common international and local formats.
- Credit card numbers: 13-19 digits with optional separators, validated with the Luhn algorithm to reduce false positives.
- ID / Passport numbers: US SSN format, Taiwan National ID format, and common letter+digit passport patterns.
- IP addresses: IPv4 and IPv6.
- Bank account / IBAN: international bank account number format.
- Manual marking: unstructured data like names can't be reliably detected by regex, so you can select text and add it manually.
Typical use case
For example, before pasting a customer support email into an AI tool for proofreading, mask the customer's name, email and phone number first. Or before asking an AI to reformat a spreadsheet containing employee data, de-identify it first. As long as the codes remain in the AI's reply, you can restore them back to real values with the mapping table — the AI only ever sees the codes, never the real personal data.
Frequently Asked Questions
Q1: Is the mapping table ever uploaded to your servers?
No. This tool runs 100% in your browser (Chrome, Firefox, Safari, etc.) — detection, masking and restoration all happen locally. We have no backend server receiving any of your text. The mapping table only exists in your current tab's memory; it disappears when you refresh or close the tab, so please download it if you want to keep it.
Q2: Why not just give the mapping table to the AI to restore it for me?
If you paste the mapping table into the AI as well, the AI would see both the codes and the real values at the same time, which completely defeats the purpose of de-identification. The restore step is deliberately designed to run entirely offline on your device so real sensitive data never passes through any third-party AI server.
Q3: Can regex-based detection miss things or over-match?
Yes — that's a known limitation of pattern-based detection. Unstructured data like names or addresses can't be reliably identified by format alone. We recommend reviewing the masked text once after detection; if something was missed, use "Mark selected text as sensitive" to add it manually so nothing important slips through.
Q4: What is PII and de-identification?
Personally Identifiable Information (PII) refers to any data that can identify a specific individual, such as a name, ID number, or contact details. De-identification replaces this data with codes or pseudonyms to reduce the risk of exposure. See Wikipedia: Personal data for more background.
You might also need