Free tool · Privacy Data anonymization: mask personal data in text and CSV files
Before a support log goes to a contractor, a spreadsheet goes into a test database or a chat transcript goes into an AI tool, take the personal data out. The tool finds emails, IPv4 and IPv6 addresses, card numbers (Luhn-checked) and Chinese mobile and ID numbers, lets you mark name, birth-date and address columns in a CSV, and applies the method you choose to each type. It all runs in your browser.
- Data stays on your device
- Checksums cut false matches
- Five methods per type
Anonymize data
Paste text or a CSV, set the rules on the right, and the result appears below.
Paste support tickets, chat logs, server logs or table data. Up to 10 MB.
Columns
Names, birth dates and addresses have no fixed format, so they are only processed in columns you mark. Other columns are scanned automatically.
Rules
| Type | Method | Keep first / last |
|---|---|---|
| 11 digits starting 13–19, +86 optional | / |
|
| Any address | / |
|
| 18 digits, valid check digit | / |
|
| 13–19 digits, passes Luhn | / |
|
| e.g. 192.0.2.45 | / |
|
| e.g. 2001:db8::1 | / |
|
| Columns you mark as name | / |
|
| Columns you mark as birth date | / |
With a salt the tool uses HMAC-SHA-256; without, plain SHA-256. Keep the salt secret and away from the data.
Result
- Mobile number (China)1
- Email1
- Card number2
- IPv4 address1
- IPv6 address1
Paste data or open a file and the result appears here.
Version v1.0Updated Update history
What each method gives you
The same email address, jane.doe@example.com, comes out differently with each method, and each suits a different job.
Mask
j*******@example.com. Part of the value stays readable for checks (“is this your card ending 0007?”); the rest becomes *. Good for screenshots, tickets and support views.
Hash
A fixed 16- to 64-character code. The same input always gives the same output, so you can join tables, de-duplicate and count people without seeing the value. Add a salt so nobody can look values up in a precomputed table.
Replace with a token
[EMAIL_1]. Each distinct value gets its own numbered token throughout the data, so an AI tool or an analyst can still tell that two messages came from the same person.
Generalise
*@example.com. IPs keep only the /24 (IPv6 /48) network, birth dates only the year, card numbers only the issuer prefix. Good for statistics by region, age group or bank.
Remove deletes the value. For whole columns you do not need, choose “Remove column”: data you do not keep cannot leak.
Anonymize data in four steps
The default rules suit sharing tickets and chat logs with a contractor or an AI tool.
-
1
Paste or open
Paste text, or choose “Table (CSV / TSV)” and paste or open a .csv file.
-
2
Mark the columns
In a table, mark name, birth-date and address columns, and remove the ones you do not need.
-
3
Set the rules
Pick mask, hash, replace, generalise or remove for each type; set how many characters masking keeps; add a salt for hashing.
-
4
Check and export
Review the highlighted preview and the counts, then copy or download the .txt or .csv.
How the tool finds personal data
It only matches formats that can be checked, preferring to miss a value over rewriting order numbers and timestamps.
- Card numbers: 13–19 digits, optionally in groups of four, and the Luhn checksum (ISO/IEC 7812-1) must pass
- Emails: a local part, @, and a domain with a top-level domain of at least two letters
- IPv4 and IPv6: every part in range; IPv6 must parse completely, so times such as 12:30:45 and MAC addresses are not matched
- Chinese ID numbers: 18 characters, a real birth date and the GB 11643-1999 check digit; Chinese mobile numbers: 11 digits starting 13–19, +86 optional
- Overlaps are resolved in this order: email › ID › mobile › card › IPv6 › IPv4, so nothing is processed twice
| Type | Default method | Example result |
|---|---|---|
| Mask, keep first character | j*******@example.com | |
| Card number | Mask, keep first and last 4 | 4111 **** **** 1111 |
| IP address | Generalise to /24 (IPv6 /48) | 192.0.2.0 |
| Name (column) | Mask, keep initials | J*** D** |
| Birth date (column) | Generalise to the year | 1988 |
| Address (column) | Mask all but last 4 | *********************isco |
| Mobile number (China) | Mask, keep first 3 and last 4 | 138****8000 |
A support ticket before and after
The ticket is processed with the default rules before it goes to an outside quality reviewer.
| Original | Anonymized | Rule |
|---|---|---|
| jane.doe@example.com | j*******@example.com | Mask the local part, keep the domain |
| card 6222 0200 0000 0007 | card 6222 **** **** 0007 | Passes Luhn; separators kept |
| from 192.0.2.45 | from 192.0.2.0 | Generalised to the /24 network |
| at 12:30:45 | at 12:30:45 | A time, not an IPv6 address |
| Ticket #48213 | Ticket #48213 | Too short to be a card number |
The name “Jane Doe” in the free text is not changed: names in running text have no reliable format. In a CSV, mark the name column instead.
Where anonymization goes wrong
The tool cleans values with a clear format; deciding whether that is enough is still up to you.
Treating hashing as anonymous
Emails and phone numbers can be guessed and hashed one by one. A hash is pseudonymisation under GDPR Article 4(5); use a secret salt and keep it apart from the data.
Leaving quasi-identifiers
Birth date, postcode and gender together identify most people. Generalise the birth date to a year and mask or remove the address.
Not checking free text
Names, addresses and licence plates inside sentences are not detected. Read the highlighted preview before you export.
Inconsistent masking
If one export keeps the first 4 digits and another the last 4, putting them together restores the number. Use one rule per dataset.
The same thinking inside the product
TapCub analytics never stores visitor IP addresses or writes cookies, encrypts contact details at rest and masks sensitive values by role.
Cookieless visitor ID
Visitors are told apart by a keyed hash of the site, a salt, the IP address and the browser type. Nothing is written to the browser and the IP is never stored.
Learn moreMasking by role
Sensitive values can be hidden or partially shown depending on who is looking; phone numbers and emails on a profile can be encrypted before storage.
Learn moreCompliance centre
Choose balanced or strict, switch on IP truncation for some regions, set retention, and look up, export or delete everything about one person.
Learn moreData anonymization questions
Still have a question?
Chat with the team behind TapCub — we usually reply within a few hours on business days.
What is data anonymization?
Changing data so it no longer points to a specific person: masking, hashing, replacing, generalising or removing names, contact details, ID and card numbers. Under GDPR Recital 26, information that is truly anonymous is no longer personal data.
What is the difference between anonymization and pseudonymisation?
Pseudonymised data can be linked back to a person with extra information, such as the original list or the salt (GDPR Article 4(5)). Hashing and tokens are pseudonymisation. Generalising and removing move you towards anonymization, but only if what is left cannot single someone out.
How do I anonymize a CSV file?
Choose “Table (CSV / TSV)”, paste or open the file, mark the name, birth-date and address columns, remove the ones you do not need, check the preview and download the new CSV.
Is my data uploaded?
No. Detection and processing run in your browser, nothing goes into the address bar and there is no share link. You can confirm it in the developer tools’ Network tab, or go offline after the page has loaded.
Does it detect US phone numbers or social security numbers?
Not automatically: phone formats vary too much to match safely in free text. In a CSV, set that column to “Mask all but last 4”, or to Remove column.
Why was a long number left unchanged?
Card numbers must pass the Luhn checksum and Chinese ID numbers need a real birth date and the right check digit; anything else, such as an order or invoice number, is left alone. ID-like numbers that fail the check are counted; tick the option below the rules to process them anyway.
Which hash does it use?
SHA-256, or HMAC-SHA-256 when you enter a salt, shortened to 16, 32 or 64 hex characters. Values are normalised first (emails lower-cased, spaces removed from numbers) so the same person hashes the same way.
Update history
New features, fixes and data refreshes for this tool, newest first.
-
· v1.0
LaunchedFirst release.
Missing something or found an error? Tell us — requests that come up often go into the next update. All tool updates
Collect less personal data to begin with
TapCub analytics tells visitors apart without cookies, never stores IP addresses, and masks sensitive fields by role.