Safe AI use
Using AI is fine. Pasting customer data into it, less so.
Your colleagues already use AI chat tools, whether you’ve arranged it or not. ClickCheck asks them to check what they’re sharing at the moment they click Send.
Can you summarise this complaint? Customer: [name], [email], [phone] Account: [account number] Order 88213 arrived two weeks late and he wants a refund.
Can you summarise this complaint? Customer: Johan de Boer, j.deboer@mail.example, +31 6 12345678 Account: NL02 RABO 0123 4567 89 Order 88213 arrived two weeks late and he wants a refund.
Sent. Nothing else happens in this demo.
Sent. Only your choice was logged, not the text.
Personal data replaced. You can send it now.
Can you summarise this complaint? Customer: Johan de Boer, j.deboer@mail.example, +31 6 12345678 Account: NL02 RABO 0123 4567 89 Order 88213 arrived two weeks late and he wants a refund.
Can you summarise this complaint? Customer: [name], [email], [phone] Account: [account number] Order 88213 arrived two weeks late and he wants a refund.
Ask anything
The risk
It rarely goes wrong out of bad intent. It goes wrong out of convenience.
Summarising a complaint, rewriting a contract, tidying up a spreadsheet. It’s handy to paste it into an AI chat. Except that email contains a name, an address and a bank account number, and that contract contains a client’s terms.
Whether and how an AI service stores or uses that data depends on the service and the subscription. Under the GDPR, an organisation has to be able to account for where personal data goes. A ban on its own rarely works; a reminder at the right moment works better.
How it works
The check runs on the person’s own computer.
That matters: a tool that checks sensitive text by first sending it somewhere else doesn’t solve the problem.
ClickCheck recognises locally
- Email addresses and phone numbers
- IBAN account numbers
- National identification numbers, such as the Dutch BSN
- Terms you mark as sensitive yourself, such as client names or project codes
And never sends on
- The text someone pastes or types
- Which data it contained
- The conversation with the AI service
Rules
How companies set it up.
-
Personal data in a public AI chat.
Warning with a choice
-
An AI service you don’t use.
Alternative: the business workspace you do pay for
-
Uploading files to an AI service.
Confirmation
-
Medical or criminal-record data.
Block
ClickCheck isn’t an AI product itself. The rule assistant helps you write rules; this warning runs on fixed recognition rules inside the browser.
Frequently asked questions
Does ClickCheck read what I type into an AI chat?
The extension checks the text locally when you send it, and only on websites that have a rule. The text itself never leaves your computer towards ClickCheck.
Does it work with every AI service?
With services you use in the browser. You set which rule applies per website.
Make a mistake once. Not twice.
Free for 3 users and 3 rules. No credit card needed.