CaraComp
CaraComp
Forensic-Grade AI Face Recognition for:
Get Started7-day refund guarantee**
Podcast

AI Deepfake Scam News Today: Fake Video Call Costs Bank €95M

AI Deepfake Scam News Today: Fake Video Call Costs Bank €95M

AI Deepfake Scam News Today: Fake Video Call Costs Bank €95M

0:00-0:00

This episode is based on our article:

Read the full article →

AI Deepfake Scam News Today: Fake Video Call Costs Bank €95M

Full Episode Transcript


In February, the chairman of an Italian private bank got a WhatsApp message. It appeared to come from the chief executive of the banking group that owns his firm. The request was urgent. Help with a deal overseas. By the time it was over, roughly ninety-five million euros had left the bank.


Reporting from Reuters puts this among the largest

Reporting from Reuters puts this among the largest known deepfake scams anywhere in the world. And the most important voice in the whole scheme wasn't real. If you've ever posted a video of yourself talking, there may already be enough of your voice online to copy it. That's an uncomfortable thought. It's also one you can do something about. The bank is Fideuram, the private banking arm of Intesa Sanpaolo, one of Italy's biggest lenders. Criminals used a fake message and a cloned voice to push money through overseas accounts. Investigators across three countries are still chasing part of it. So if a familiar voice can't prove who someone is anymore, what can?

Start with the man at the center. His name is Paolo Molesini, and he chairs Fideuram. According to reports, the first message looked like it came from Carlo Messina, the group's chief executive. Then came a phone call. The caller claimed to be a senior partner at a law firm. A.I. software had copied that lawyer's voice. A text from the boss. A call from the lawyer. Neither one was real.

How hard is it to fake a voice like that? Easier than most people assume. Security researchers say about a minute of recorded speech can be enough. Scammers pull clips from interviews, speeches, or social media. They feed those clips into an A.I. tool. Then they type whatever they want said. The software speaks it back with the person's tone, accent, and rhythm. Researchers describe this as crossing the "indistinguishable threshold." In plain terms, most listeners can no longer tell the copy from the original. For banks, that breaks a decades-old habit of trusting a call from someone you know. For the rest of us, it means a panicked call that sounds like your son or your mother might not be them.


Trusted by Investigators Worldwide
Run Forensic-Grade Comparisons in Seconds
Detailed facial comparison reports. Results in seconds.
Get Started
7-day refund guarantee**

This wasn't a one-off, either

This wasn't a one-off, either. In 2025, Italian businessman Massimo Moratti sent nearly one million euros to fraudsters. They'd used a cloned voice to pose as Italy's defense minister, Guido Crosetto. And the wider numbers point the same direction. According to Keepnet Labs and the research firm Gartner, deepfake attacks that fool Instagram Verification How A Paid Badge Hid An 8m Scam or voice security checks grew roughly eightfold in 2023. Gartner also predicted that by this year, nearly a third of companies wouldn't trust standalone identity checks on their own. That's a big shift for the tools that guard your bank account. It means the voice login you set up years ago is no longer the whole lock.

Now for the part of this story that went right. Authorities in Italy, China, and Portugal worked together to trace the money. They got back about fifty-three million euros. That's more than half. Roughly thirty-six million still hasn't been traced. Reports say it moved through overseas accounts and was turned into cryptocurrency. So why did so much come back? The bank's systems flagged the transfers, and staff moved fast through international channels. Security analysts say that speed came from a plan written in advance. Contacts at other banks. Clear steps for who to call and when. For fraud teams, the sharpest lesson is simple. Require a second named approver inside the system for any payment above a set amount. Then no single cloned call can move the money alone.

Critics worry that rules like that slow down honest business. But in this case, the bank's quick response recovered the money, not its approval process. Automatic checks, like forced callbacks and multiple approvers, don't ask a stressed person to make the right call under pressure.


The Bottom Line

The bank never caught the fake voice. It caught the money. What saved tens of millions wasn't sharper ears. It was a system built on the assumption that someone, someday, would be fooled.

Scammers copied a trusted voice and tricked a bank chairman into sending about ninety-five million euros. Nobody could hear that the voice was fake, because today's A.I. makes that nearly impossible. The bank got more than half back because it had a plan ready before anything went wrong. Whether you're approving a wire transfer or answering a frantic call from family, the safest move is the same. Hang up, and call back on a number you already trust.

Full breakdown's in the show notes.

Ready for forensic-grade facial comparison?

Full forensic reports with detailed similarity scoring. Results in seconds.

Run My First Search