The AI doesn't need to know who

A customer support agent gets an angry email at four in the afternoon. A refund went wrong, the customer is upset, and the reply needs to be calm, clear, and kind.

So he does what millions of people now do every day. He copies the whole email, pastes it into an AI chat, and types: "Help me write a polite reply."

Thirty seconds later, he has a great draft. Problem solved.

Except that email had the customer's full name, phone number, bank account number, and the details of a transaction. All of it is now sitting in a chat history on someone else's server. The customer never agreed to that. The agent didn't even think about it.

He wasn't careless. He was just trying to do a good job, fast.

It happens everywhere

This isn't only a support problem. An HR manager pastes a performance review to make the wording kinder. An accountant pastes a spreadsheet to find a mistake. A lawyer pastes a contract to summarize it. Someone pastes a family member's medical results to understand them.

Every one of those is a reasonable thing to want help with. And every one of them hands over details about a person who isn't in the room.

That's the part that bothers me. Your own data is your choice. But a customer's name, a colleague's salary, your mother's diagnosis? That information belongs to them. They trusted you with it, not your AI tool.

Privacy is about respect

Privacy often sounds like a legal topic, full of policies and checkboxes. I think it's simpler than that. It's about respect for people who aren't there to speak for themselves.

Once you paste something, it's out of your hands. Depending on the tool and its settings, that conversation may be stored, reviewed, or used to improve the service. It may live in your account for years, on every device you're signed in to. And if that account is ever compromised, everything you pasted goes with it.

Most of the time, nothing bad happens. But "probably fine" is not how you'd want someone else to treat your bank account or your health records.

It needs the situation, not the person

Here's the thing most people miss. To write that reply, the AI doesn't need the customer's real name. It doesn't need the phone number or the account number. It needs the situation: a refund failed, the customer is frustrated, the tone should be warm.

"Customer A" works just as well as a real name. "Account X" works just as well as a real number. The quality of the answer barely changes. The risk changes a lot.

So before you paste, take ten seconds:

• Replace names with placeholders like Customer A or Employee B.
• Remove numbers that identify someone: phone, ID, bank account, card.
• Keep the context the AI actually needs to help.
• Put the real details back yourself, after you get the answer.

And check your company's rules. Many companies already have an AI policy. Many employees have never read it.

Convenience shouldn't cost someone else

AI is becoming part of how we write, think, and work. That's a good thing, and I use it every day. But the faster it gets, the easier it is to forget whose information we're moving around.

Protecting data used to be a job for the security team. Now it's a habit for everyone who has a chat box open. I wrote about why leaky data is worse than messy code. This is the everyday version of the same idea.

Use AI as much as you want. Just let it help with the problem, not learn who the problem is about.