Don't tap the link. To check a suspicious text, copy its wording, paste it into ChatGPT, and ask for the warning signs. ChatGPT helps you understand why the message looks off; the decision about what to do next always stays yours. The check takes about a minute, works on the free version, and you never touch anything inside the message itself.
Scam texts are big business. The FTC's data spotlight on text scams put reported losses at $470 million in 2024, with fake package-delivery texts among the most reported. The messages look cleaner than they used to, because scammers now use AI to write them. The good news: you can use the same technology, calmly and for free, to take them apart.
How do I check a suspicious text with ChatGPT?
You copy the wording of the message, paste it into ChatGPT with one question, and read the warning signs it names. You never tap the link, never reply, and never call a number from the message. Here is the whole method, step by step.
- Leave the text alone. Don't tap, don't reply, don't call. A text just sitting in your messages can't hurt you.
- Copy the wording. Press and hold the message bubble and choose Copy. On most phones a small menu appears right over the message.
- Open ChatGPT. Go to chatgpt.com or open the app. You'll see a plain screen with a message box at the bottom. Never used it? ChatGPT for Beginners walks you through the free signup.
- Paste the message and ask. Tap the message box, paste, and add the question below. Press Enter or tap the arrow, and the answer appears in a few seconds.
- Read for warning signs, not a verdict. ChatGPT names what looks off, point by point. Your job is to understand the signs, not to obey a yes or no.
Here is the exact question. Copy it word for word and replace only the part in parentheses:
I received this text message and I'm not sure it's real. Please list the warning signs that suggest a scam, explain each one in plain English, and tell me what I should check for myself before doing anything. Here is the message: (paste the message here)
One safety note: paste only the wording of the message. Never type your own passwords, verification codes, or account numbers into the chat. The check doesn't need them.
What does a typical scam text look like?
Most scam texts imitate a delivery company, a toll agency, or a bank, and nearly all of them push a link with a deadline. Here is a made-up example of the kind the FTC says gets reported most, the package text:
- 1The made-up deadline. Real companies don't pressure you like this.
- 2The link doesn't match the company's real website.
And here is what happens when you run the check on it:
- 1Your question, with the suspicious text pasted below it
- 2ChatGPT names each warning sign and explains it
What warning signs will ChatGPT point out?
Four tells come up again and again: manufactured urgency, a link that doesn't quite match the real company, a request for codes or personal details, and an odd payment method. This kind of fake text even has a name: smishing, which is phishing carried out by text message.
- Urgency. "Within 12 hours," "immediately," "final notice." Scammers create urgency. Real organizations don't.
- A link you didn't ask for. Often one that resembles a real company's name but isn't quite right.
- A request for information. Codes, passwords, card numbers, your Social Security number. No real company asks for these by text.
- A strange payment method. No real business or government agency will ever tell you to pay with gift cards.
What about the Malwarebytes and Norton scam checkers inside ChatGPT?
They exist, they're free, and they're worth knowing about. Since early 2026, the security companies Malwarebytes and Norton have offered official scam-checking apps that run inside ChatGPT: you type @Malwarebytes in the message box, paste the text, and ask whether it's a scam. That's a quick second opinion, and using it is perfectly sensible. The reason we still teach the paste-and-ask method is that it teaches you the signs. A checker gives you an answer about this message; understanding urgency, mismatched links, and information requests protects you on the next one too. And no checker, ChatGPT included, can see who really sent a message. The final call stays with you either way.
What can ChatGPT get wrong?
ChatGPT reads only the words you paste. It can't see who really sent the message, can't check whether you actually have a package or an unpaid bill, and can't look inside a link. Sometimes it hedges, and once in a while a genuine message will trip the same alarms as a fake one. So use it for what it's good at, explaining the warning signs, and keep the one rule that covers every gap: when money or accounts are involved, verify with the company directly, through a phone number or website you look up yourself. The same paste-and-ask method works for email too, and Can ChatGPT Tell if an Email Is a Scam? covers what's different there.
What if I already tapped the link?
Stay calm; a tap by itself is usually not a disaster. Don't type anything on the page that opened, and close it. If you entered a password anywhere, change that password now. If you gave card or bank details, call your bank at the number on the back of your card, never a number from the text. Then report the message: forward it to 7726 (it spells SPAM), the free reporting shortcut the FCC recommends, report it at ReportFraud.ftc.gov, and delete it. If you'd rather talk it through with a person, the AARP Fraud Watch Network runs a free helpline at 877-908-3360.
What about the scams that come by phone call?
Texts are only one channel. The same criminals also call, now sometimes with AI-copied voices. Two guides cover that side: the family safe word, a simple system that stops impostor calls cold, and how AI voice cloning scams work. If you're reading this because you worry about a parent, start with how to help your parents avoid scams. And if a parent has met an online "sweetheart" who needs money, see romance scam warning signs. Every guide lives in our scam protection library.
The five rules to keep
- Never tap links in texts you didn't expect.
- Never share passwords, codes, or card numbers with someone who contacted you first.
- Real organizations don't pressure you. Scammers create urgency.
- Verify with the company through a number or website you look up yourself.
- When in doubt, paste the wording into ChatGPT and read the warning signs before you act.
If you'd like to practice this calmly, on realistic examples and with someone walking beside you, the course AI for Real Life does exactly that: checking texts, emails, and calls step by step, at your own pace.