Does it upload my text?
No. The Text Anonymizer runs entirely in your browser, so the anonymization stays local and your text is never sent to a server.
What does it mask?
It detects and replaces common sensitive values such as emails, phone numbers, URLs, IP addresses, and long IDs with neutral placeholders.
Does it detect names or use AI?
No. This version is rule-based only and does not attempt person-name detection, which keeps results predictable and fully local.
Is it free to use?
Yes. The Text Anonymizer is free with no signup and works fully in your browser, even offline once the page has loaded.
Does it use AI, and does the text leave my browser?
No to both. The tool applies a small set of regular expressions in the page. Nothing is sent to a model or a server, which is what makes it safe to paste a real log into it. The trade-off is that it finds patterns, not meaning.
Why is the name in my text still there?
Because a name has no shape a regular expression can recognise. Emails, IPs, URLs and long numbers all have a fixed structure, so they can be matched reliably. A person name is only words, and matching it would need a language model with the false-positive problems that brings. Anything name-shaped has to be removed by hand.
What exactly gets replaced?
URLs become [url], email addresses become [email], phone-shaped digit runs become [phone], IPv4 addresses become [ip], runs of 7 or more digits become [id], and strings of 24 or more characters from the set of letters, digits, underscores and hyphens become [id] as well. Everything else is left as it was.
Can I get the original text back from the output?
No. The replacement is destructive and there is no mapping table. Every email collapses to the same [email] placeholder, so two different addresses become indistinguishable. Keep the original if you need it, and treat the output as a one-way export.
Will it catch a credit card number?
A 16-digit card number written as a single run of digits, or with spaces or hyphens between the groups, matches the phone pattern or the long-ID pattern and gets replaced. That is a side effect, not a designed feature. There is no Luhn check, no card-brand detection, and a card written in an unusual format may survive.
Is this enough before pasting a log into a public issue tracker?
It is a good first pass and a bad last one. It removes the identifiers that leak most often by accident: emails, internal URLs, IPs, tokens and long IDs. It does not remove names, internal hostnames, file paths or free-text descriptions of your customers. Read the output before you publish it.