AI is giving scammers the tools to build trust over weeks or months before striking. Here, Juliet Moran, Technical Director at TelephoneSystems.Cloud, explores how sophisticated fraud is evolving and why independent verification is becoming more important than ever.
“Most of us think we can recognise a scam call. Someone asks for bank details, passwords or security codes, usually accompanied by urgency and pressure. But what happens when none of those warning signs appear?
That was the experience of a tech-savvy AI expert who recently came dangerously close to falling for a sophisticated fraud.
It began with a telephone call supposedly from the police. The caller explained that his personal information had been discovered during a criminal investigation and provided a crime reference number.
Naturally, he searched the telephone number on Google. Several results appeared to confirm that it belonged to the police department. More importantly, the caller asked for nothing. No passwords, personal information or money. Just a professional conversation.
Some time later, another supposed officer called. Again, the conversation was reassuring. They discussed sensible precautions and what to watch for if his information had been compromised.
Then came a third call, this time discussing cryptocurrency security. The caller recommended installing a genuine security application. Nothing seemed particularly suspicious until they suggested downloading it from a specific website rather than the app store.
That website was the trap.
The criminals had created a convincing imitation designed to compromise his cryptocurrency. Fortunately, the unusual domain extension caught his attention.
What makes this attack so concerning is not the fake website, but the patience involved.
Traditionally, scammers needed to move quickly. Every unsuccessful telephone call cost time and money. That encouraged pressure tactics, urgency and demands for immediate action.
AI fundamentally changes that equation.
AI-powered systems can potentially manage thousands of conversations simultaneously, remembering previous discussions and maintaining a consistent story. They do not become impatient, and the marginal cost of another conversation can be extremely low.
The first call might simply establish trust. The second reinforces credibility. The third offers genuinely useful advice. Weeks or months later, the real scam begins, when the caller no longer feels like a stranger.
Criminals can also manipulate the information victims use to verify them. AI makes creating convincing fake websites faster and cheaper, while techniques such as SEO poisoning can place misleading information in search results.
A Google search showing several websites associating a telephone number with the police does not necessarily prove anything. Those websites can be created specifically to support the deception.
Voice cloning creates another threat. AI-generated voices can imitate colleagues, managers or family members, making even familiar voices unreliable as proof of identity.
The defence is independent verification. If the police call, contact them through an independently verified number. If someone recommends software, find the official website yourself. For financial transactions, establish a separate verification process, potentially including an agreed safe word.
The frightening thing about AI-enabled fraud is not simply convincing voices or fake websites. It is that criminals can now automate something that previously required considerable human effort: building relationships over time.
The first call does not need to steal anything. It only needs to make you trust the second.
And by the time the real request arrives, you may have forgotten why you should be suspicious.
That is what makes the AI-powered long con so dangerous.”
