AI privacy filter

Ask the AI. Keep the names to yourself.

Paste an e-mail, a chat or a document and every name, e-mail address, phone number, bank account and address becomes a placeholder like [NAME_1]. Paste that into ChatGPT, Claude or whatever you use. When the answer comes back, paste it here and the real names return. The chatbot never saw them.

1. Your text

2. Safe to paste

The version without personal details shows up here.

Copy with a note asking the AI to keep the placeholders

Tip: click any word above to hide it everywhere, click a coloured tag to show it again.

3. Got an answer? Put the real names back.

Paste the AI's reply here and every [NAME_1] turns back into the real thing. The list of what's what only lives in this tab: close it and it's gone.

The answer with the real details shows up here.

Questions, probably

Why not just paste it into the AI?: Whatever you paste into a chatbot is stored on someone's server, may be read by staff for quality checks, and depending on your settings may be used to train future models. Your clients, colleagues and family didn't sign up for that.
Does this use AI to find the details?: No, and that's the point. It uses patterns and checksums (the same maths banks use to validate an IBAN or card number) right here in your browser. Nothing is sent anywhere, not even to me.
Will it catch everything?: E-mail addresses, phone numbers, IBANs, card numbers and ID numbers, pretty reliably. Names are harder: I find them after greetings and titles, in signatures, e-mail headers and e-mail addresses. Anything I miss, click it once and it's hidden everywhere. Always have a quick look before you paste.
Is it GDPR-proof?: It helps a lot, but it's not a legal stamp. Replacing names with placeholders is pseudonymisation: the details are gone from what you paste, but the context can still say something. For really sensitive stuff, ask yourself if it should go into an AI at all.