Copy article

AI scams are on the rise – what do you need to ask it to reveal that it is AI?

ended 13. April 2026

AI scams are on the rise with many phone calls and text messages now generated by AI. Many are indistinguishable from those created by humans.

  • What do you need to ask it to reveal that is is AI-generated?
  • How big a problem are scams created by AI?
  • What other advice do you have?

Responses asap.

4 responses from the Newspage community

Copy all

Copy

Editor
Katrina Young
Futurist | Chief Technology Officer CTO/CDO Ethical AI & Digital Transformation Strategist at KYC Digital
AI-powered scams are no longer a fringe risk. The same technology that makes AI useful — voice synthesis, personalisation, real-time responsiveness — makes it an effective tool for fraud.

The instinct most people follow is wrong. Asking “are you AI?” does not work. A well-configured scam system will deny it. The better approach is a pattern interrupt: say something unexpected, pause mid-sentence, give a false name or detail. Legitimate systems handle disruption. Scam systems often don’t.

The rules are simple but not obvious. Never confirm personal data on an inbound call or message, even when the caller already appears to know it. That’s the hook, not the proof. If something feels rehearsed or pressured, it probably is.
Protective tools now exist that do the heavy lifting — software like Norton 360 can flag robo-calls, scan incoming messages, and identify phishing attempts before you engage. I built Digital Guardian specifically for this gap: consumers need a protective AI companion.
Copy

Ben Foster
CEO at The SEO Works
Always have a second way to check who you are talking to. If you get a worrying call or text, hang up and use a trusted number or a different app to reach that person or company directly. Never feel pressured to act quickly, as that is exactly what the scammer wants you to do.

To catch out a machine, you need to move away from a normal conversation. You could try the Car Wash test that has been proven to trip up AI. Ask "I want to wash my car. The car wash is 50 metres away. Should I walk or drive?". Most AIs recommend walking due to the environmental benefits (which would leave the car at home).
Copy

Rohit Parmar-Mistry
Founder at Pattrn Data
AI scams are becoming dangerous because they remove the old friction points people relied on. You used to spot the bad grammar, the odd phrasing or the obvious copy and think twice. Now a scam text, email or voice note can sound polished, personal and completely plausible.

The first thing to ask is simple: who are you, what organisation are you calling from, and what specific action do you want me to take right now? Then slow it down. Ask for a reference number, ask them to email from an official domain, or say you will call the company back through the number on its public website. Most scams fall apart when you take the timing away from the scammer.

The bigger problem is volume. AI lets fraudsters personalise at scale, so the con feels more believable and lands more often. My advice is blunt: never trust urgency, never trust a voice just because it sounds human, and never approve money, passwords or codes without an independent check.
Copy

Colette Mason
AI Ethics Consultant at Clever Clogs AI
The real question is why banks still treat a voice as proof you said yes.

National Trading Standards exposed criminals using "lifestyle survey" calls to harvest voice samples, then cloning those voices to authorise direct debits. Three seconds of audio is enough. The tools cost less than a streaming subscription. NTS blocked nearly 21 million scam calls in six months. 73% of UK adults were targeted last year, and an estimated 19 million lost money.

Listen for awkward pauses, odd reactions to unexpected questions, robotic word stress on phishing calls. Agree a safe phrase with family.

The problem is detection advice puts the burden on the person least equipped, often elderly, often alone, while financial systems still accept voice consent as authentication in a world where a voice proves nothing.

If your voice can be faked in seconds, voice-based consent isn't consent anymore and financial institutions need to recognise this and swiftly adapt