unpii.me

Removes personally identifiable information from any text.

Nothing stored - runs entirely on our servers in Germany, no logging of your text, no training on your data.

Paste text here
DE-1·DE-2·DE-GPU

Catches most personal data - not all of it. Review your output before sharing.

Anonymized result
Marker format
DemoThis is what your result looks like from the Docs plan up - a real model result on the sample text.
Hi team, from Acme Corp here. Meeting on . Email: , phone . Office: . Project page: IBAN , ref . Portal token: . PS: john will call again tomorrow about the access data.

Everything here is interactive: hover markers for their confidence, click the red dashed uncertain hit to redact it - and if something is missing, just select it yourself. See plans

Best-effort PII detection: the model is trained on synthetic data and covers multiple languages.

Anonymized in Germany · EU servers · Privacy by design

See all features →

What is PII?

PII stands for Personally Identifiable Information. It covers any data that can identify a person directly or indirectly: name, email, phone, address, date of birth, IBAN, IP address - sometimes only in combination.

How it works

A local AI model with 1.5 billion parameters runs on our server in Germany. It reads your text, detects personal data, and replaces it with placeholders like [PERSON_1] or [EMAIL_1]. Your text stays in memory only and is discarded right after the response. Nothing is stored here.

Who is it for

For anyone who needs to quickly remove personal data from their texts and documents. Whether HR, lawyer, compliance officer, or developer sanitizing their LLM input. For a fairer way to share data.

If you're going to send data to the US anyway, at least redact it at home.

What gets detected

[PERSON][EMAIL][PHONE][ADDRESS][DATE][URL][ACCOUNT][SECRET]

The more context, the better the detection: a standalone first name scores lower confidence than a full name in a sentence.

Please keep in mind: we're not perfect either, even though we're giving it our all. Double-check your texts before sharing them.