Our neighbor Dot is a smart cookie. Seventy-four, sharp as a tack, and she's been using a chatbot on her phone for about a year. When an email arrived saying her bank account was "temporarily restricted," she did what she thought was the careful thing. She pasted it into the AI and asked, "Is this a real email from my bank?"
The answer came back sounding very sure of itself. The email "appears professionally formatted," it said, and "includes standard security language." It suggested she follow the link to verify her account. Dot's finger was on the link.
Well, folks, what saved her was an old habit. Before tapping, she opened her actual bank app, the one she always uses. No restriction. No message. Nothing. The email was a fake, and the link led to a copycat login page.
Can AI tell me if a message is a scam?
AI can help, but it can't give you a guarantee. A chatbot can point out common warning signs, like pressure to act fast or a strange web address. What it can't do is check with your bank, see your real account, or know for sure whether a link is safe. It only sees the words you paste in.
And AI can simply be wrong while sounding confident. Google says plainly that AI responses can be inaccurate. Scammers also write well now, so "looks professional" means nothing.
How should I use AI on a suspicious message?
Here are the four steps in plain words. First, remove your name, account numbers, and any links before you paste the message in. Second, ask a better question. Third, never click a link just because the AI said it looks safe. Fourth, check with the company yourself, through your own app or a phone number you already have saved.
That second step is the big one. Don't ask "Is this real?" Ask this instead, and feel free to copy it: "What parts of this message would I need to check with the company myself before I do anything?" That turns the AI into a helper that gives you a checklist, not a judge that gives you a verdict.
What should I look for no matter what the AI says?
Look at what the message wants you to do. The FTC's guide on how to avoid a scam describes the same pattern over and over: someone pretends to be a company you know, says there's a problem, and pushes you to act right now.
In plain words, stop if a message asks you to click a link to fix your account, share a code that was texted to your phone, pay with gift cards, a wire transfer, or cryptocurrency, or install an app so someone can "help" you. Those requests are the scam, no matter how nice the email looks.
What if the AI says it's definitely a scam?
That's a helpful warning, and you should still check through your own app or saved number. Once in a while, a real company does need you, like a genuine bill that's overdue. Checking yourself keeps you from both panicking and ignoring something real.
AI is still a wonderful tool for everyday paperwork. Bob's guide on using AI to explain a confusing letter shows a safer way to put it to work. And if you did click a bad link, here's what to do next.
What Dot tells her book club now
Dot still uses her chatbot almost every day, for recipes, crossword clues, and letters to her congressman. But she's changed one thing. "I ask it what to check," she told her book club, "and then I go check it myself." Nobody in that book club has clicked a bank link since.
The AI can be a good second opinion, folks. Your bank's own app is the only first opinion that counts.
Have you ever asked an AI whether a message was a scam? Tell us in the comments what it said.
