You see a message from your mom: “The bank says I owe $500, here’s the proof.” She attached a picture that looks like a bank statement, but it was made by an AI. You feel a pang of worry.

You’ve been there before — an AI chatbot gave a support ticket number that turned out to be fake, and nobody noticed for 3 days. The confidence in the answer made it feel real.

That’s the problem: AI sounds sure even when it’s wrong. Confidence is not proof. If you act on it without checking, you could move money, change a medication, or miss a real issue.

The simple test is to ask “how do you know that?” If the AI can’t point to a source, treat it as a guess.

Treat every AI answer as a draft, not a final decision

AI is good at pulling together words fast. It can give you an outline for a letter, a rough idea for a gift, or a summary of a long article. That’s useful. What it isn’t good at is guaranteeing that every fact it states is true. Think of the output as a first draft that needs a human editor. You wouldn’t send a rough draft to your boss without reading it, and you shouldn’t act on an AI answer without checking it.

The ticket-number trick: why fake details feel real

A chatbot once gave a customer a ticket number that did not exist. Nobody noticed for 3 days because the number looked plausible and the chatbot sounded certain. When you see a specific number — ticket, reference, account — your brain treats it as evidence. AI knows how to generate those details, but it doesn’t have a database to pull them from. It’s making them up on the spot, and the confidence tricks you into believing they’re real.

Photos lie: AI-generated images are not evidence

People forward AI-made photos as if they were real evidence. A picture made by AI proves nothing about the world outside the model. If your parent shows you an AI‑generated “bank statement” or a “doctor’s note,” you need to treat it like a rumor, not a document. The only way to know if it’s real is to go to the source — log in to the bank’s website, call the clinic, or ask for a paper copy.

When speed feels like certainty: the instant answer trap

If an AI answers instantly with no source, treat it as a guess, not an answer. Speed is not a sign of accuracy; it’s a sign that the model is pulling from patterns it has seen before. Real verification takes time — opening a website, reading a bill, speaking to a person. When the AI skips that step, you should too.

Your verification checklist: bill, account, real person

My rule: for anything important (money, health, work), I check the original source myself — the bill, the account, the real person. That means:

  • Money: log in to the bank or credit‑card site, look at the transaction, or call the number on the back of the card.
  • Health: check the prescription label, call the pharmacy, or ask the doctor’s office directly.
  • Work: open the ticketing system, look at the email thread, or speak to the colleague who sent the request.

If the AI’s answer doesn’t match what you see at the source, ignore the AI and follow the verified information.

The “how do you know that?” test in practice

Next time an AI gives you a fact, ask out loud (or in your head) “how do you know that?” Listen for a source: a website, a document, a person’s name. If the answer is “I read it somewhere” or “it’s common knowledge,” treat it as unverified. If it can point to a specific, checkable source, go verify that source before you act. This habit takes 10 seconds and saves a lot of trouble later.

Keep yourself the checkpoint, not the police

You don’t need to audit every message your parent receives. Give them one simple rule instead: if something feels urgent or asks for money, pause and show it to you. You can then run the verification steps above together. If your parents have already discovered ChatGPT, it’s worth reading how their prompts can become a privacy problem — pasting bank statements into a chatbot is exactly the kind of thing the pause-and-show rule prevents. Over time they’ll learn to spot the signs — urgent language, unexpected attachments, too-perfect photos — and you’ll both be less likely to act on a false alarm.

Treat AI like a helpful intern: fast, eager, but always needing your sign-off before anything goes live.

What’s the one thing you make them show you before they click? Tell me and I’ll add it to the checklist.

Related reads: