A familiar voice does not always mean a familiar person
You receive a message or call that sounds convincingly like a loved one or a familiar company, asking you to act immediately, for example by sending money or a code. AI can create a convincing image, video or voice message, so a familiar voice or image alone is not proof of who sent it. Urgency, emotion and pressure are tools of a scam: they make you act before you have time to check.
Make your first check
- 1
Stop
Do not send money, authentication codes or images in a rush. A scam often tries to prevent you from asking for advice.
- 2
Check what is being requested
Ask yourself whether you are being asked for money, a password, a verification code, a gift card or something else that would not normally be requested this way.
- 3
Verify through another channel
Contact the person, company or authority through an official channel you find yourself. Do not rely on contact details in the message itself.
- 4
Ask another person
If you are unsure, show the message to someone you trust before acting. Embarrassment or urgency are no reason to handle it alone.
Try this checking request
A checklist for a possible scam message
Make a checklist for this fictional message. It urgently requests money and says not to tell anyone else. Separate the message's claims, pressure tactics and things I will verify myself using an official source or another contact channel. Do not infer the sender's identity from the message. Message: "Hi, I need your help right now, but this is confidential, so do not mention it to anyone. Buy four 100-euro gift cards and send pictures of the codes to this number; I will invoice the amount later. I cannot call, I am in a meeting."
Compatible with: AI assistant
Continue in the same conversation: what to say when I call to check
Use the checklist you just gave me. Add a separate section at the end: what I should say when I call to check, without revealing details of the message before I know who is answering. Keep the rest of the checklist unchanged.
What checking looks like in practice
Example prompt:
You receive a voice message that sounds like a loved one, asking you to buy a gift card immediately and keep it secret.
Result:
I will not rush to buy a gift card or send codes. I will call my loved one on a number I saved myself and ask directly about the situation. If the message turns out to be a scam, I will report it following the service's instructions.
From an emotional reaction to a checklist
A common request:
Is this a real emergency?
Try asking this instead:
Make a calm checklist for this message. Identify urgency, secrecy, the request for money and new contact details. Tell me what to verify through another channel before replying. Do not infer who sent the message. Message: "Hi, this is a new number; my old one was stolen. I need money from my account for a transfer right now, but do not mention this to anyone until I have had time to explain myself. Send 250 euros to this account number as soon as possible; I will pay it back as soon as I can get to the bank."
A good request helps separate the message's content from things to check, rather than asking AI to try to prove the sender's identity.
Remember these warning signs
- Urgency: you must act immediately.
- Secrecy: you are told not to tell anyone else.
- An unusual payment: you are asked for a gift card, cryptocurrency or an authentication code.
- A new contact method: you are directed to a number, email address or link you do not recognise yourself.
- Strong emotions: fear, guilt or excitement drives the decision before you check.
If you have already acted
Immediately contact the bank or service used for the payment or authentication. Change any compromised password, report the scam and, if necessary, seek help from the police or the guidance of Finland's National Cyber Security Centre.
Once you know the basics
Agree in advance with your family or colleagues on a shared safety word or question that an outsider would not know. It makes verification quicker if the situation ever arises.
Free practice on this website
Verify a request through a channel you choose yourself
Write your own answer in each round before opening the hint or checking criteria; they appear only after your first attempt. The material stays visible throughout the exercise, so you can return to it at any time. The message is made up: use it as provided, rather than your own real situation. Do not act on the response alone; always verify through a separate channel you already know.
The answers you write stay only in this browser tab. They are not sent to AI Licenses' server or to analytics.
Question 1/2
Question 1/2
You are a mother who receives this message, supposedly from her child: "Mum, my phone broke; this is a new number. I need 400 euros today for a bill. Do not tell Dad; I will explain tonight." Write what you cannot know for certain from this message and what you will do next before sending anything.
Use the practice activity in your own AI service (optional)
Rather than asking whether it is genuine, ask what you can check.
AI cannot know who sent a message. It can still help you identify what can be verified and in what order.
Verification exercise · about 5 min
Where shall we start?
How the request is built
- Task
- Explain what I cannot infer and what I should do
- Background
- The message is fictional and provided for the exercise
- Format
- Actions in order
Below is a fictional message. First tell me what you cannot infer from it. Then list, in order, what I should do next and what I should verify through another contact channel. Do not judge who sent it or whether it is genuine. Message: "Mum, my phone broke; this is a new number. I need 400 euros today for a bill. Do not tell Dad; I will explain tonight."
Ready to try it in the actual tool?
The practice message is made up. Use it as provided, rather than your own message.
A rule to stick to: Do not act on the response alone. Always verify through another contact channel: call a number you already had.
Skills and concepts in this guide
Publication details and sources
- Published:
- August 13, 2026
- Sources checked:
- September 4, 2026