What kinds of PII does this detect?

It detects email addresses, phone numbers in common international formats, credit and debit card numbers, IPv4 addresses, URLs, dates of birth, and basic street addresses. Pattern matching is conservative, so always review the output before sharing.

Does it use consistent placeholders?

Yes. The same value is always replaced with the same token within a single pass — the first email becomes [EMAIL_1] everywhere it appears — so the AI can still reason about relationships without seeing the real data.

Can it catch every name?

No, and no client-side tool can. Personal names have no reliable pattern, so this tool focuses on structured identifiers it can match accurately. Redact free-text names manually before relying on the output for sensitive material.

Is anything uploaded?

No. All detection and replacement happens locally in your browser using regular expressions. Nothing is sent to a server, logged, or stored.

Is this enough for regulatory compliance?

It is a helpful first pass, not a compliance guarantee. For GDPR, HIPAA, or similar obligations, combine automated redaction with human review and a documented process.

What is the PII Redactor for AI Prompts?

Paste your text and instantly redact personally identifiable information — email addresses, phone numbers, physical addresses, national ID numbers, and credit cards — replacing each with a consistent placeholder token. Runs entirely in your browser; nothing is uploaded. It runs free in your browser on Gera Tools, with nothing uploaded.

PII Redactor for AI Prompts

Name: PII Redactor for AI Prompts
Creator: Gera Tools
License: https://creativecommons.org/licenses/by/4.0/

Pasting raw text into an AI assistant is convenient, but customer emails, support tickets, and logs are full of personal data you should not be sending to a third party. This PII redactor scans your text in the browser and replaces structured identifiers — emails, phone numbers, card numbers, IP addresses, URLs, dates of birth, and street addresses — with consistent placeholder tokens, so you can get the help you need without leaking anyone’s details.

What it detects and replaces

The tool covers the most common structured PII patterns:

Type	Example	Token
Email address	[email protected]	`[EMAIL_1]`
Phone number	+44 7700 900123	`[PHONE_1]`
Credit/debit card	4111 1111 1111 1111	`[CARD_1]`
IPv4 address	192.168.1.45	`[IP_1]`
URL	https://example.com/account	`[URL_1]`
Date of birth	14/03/1985	`[DOB_1]`
Street address	12 Baker Street, London	`[ADDRESS_1]`

Names are intentionally excluded. Personal names have no reliable pattern across languages and cultures, and false positives would redact genuine content. Always scan the output yourself for names before pasting into an AI.

How it works

The tool runs conservative regular expressions over your text, one per identifier type. Each unique value is assigned a stable token on first appearance and that same token replaces every subsequent occurrence of the same value within the document. This preserves relationships in the text — the AI can still see that the same email address appears in two places — while hiding the actual data.

Detection runs entirely client-side. There is no network call, no logging, and no storage, so even the original unredacted text never leaves your machine.

Worked example

Input text:

Hi team, Jane ([email protected], 07700 900123) placed order #4421.
Please cc [email protected] on the response.

After redaction:

Hi team, Jane ([EMAIL_1], [PHONE_1]) placed order #4421.
Please cc [EMAIL_1] on the response.

The same email maps to [EMAIL_1] in both positions, so the AI can still reason about the relationship. The order number is preserved because it is not a PII pattern.

Using redacted output effectively

Keep the tokens in your prompt and ask the AI to use them in its response. When you receive the answer, re-insert the real values locally before acting. This workflow means sensitive data never leaves your machine, while the AI gets enough context to help correctly.

For regulated data under GDPR, HIPAA, or similar obligations, treat this as a useful first pass and combine it with human review and a documented process — automated redaction alone is not a compliance programme.

What this tool cannot detect

Pattern-based redaction has inherent limits:

Personal names — no reliable pattern; must be removed by hand.
Account numbers, policy numbers, case IDs — too organisation-specific for a generic regex.
Indirect identifiers — combinations of non-PII fields that together identify a person (for example, a rare job title in a small town). These require contextual human judgement.
Images and PDFs — the tool processes plain text only.

Use this tool for structured identifiers in text logs, emails, and support tickets. For documents with unstructured personal information, supplement with manual review.