Deepfake Scams: 1 in 5 in Hong Kong Trust Banks to Stop Them

An employee at the engineering firm Arup joined a video call with the company's senior executives. Every face on that call was fake. The employee then sent twenty-five million dollars to the fraudsters.

If you've ever taken a video call from someone you

If you've ever taken a video call from someone you trust, this story is about you. Most of us believe we'd know if a familiar face or voice were fake. That belief is exactly what these scammers count on. It's an unsettling thought, and it's reasonable to feel unsettled by it. A recent survey in Hong Kong, reported by Asian Banking and Finance, found that only about one in five people trust their banks to stop deepfake scams. A deepfake is a fake video or voice, made by A.I., that copies a real person. So if our eyes and ears can't be trusted, what can?

Start with the Arup case. That was early in 01/2024. The employee wasn't careless. He saw what looked like his bosses, talking to him live. His senses said yes. No one made him stop and check.

Now the second case, from February of this year. According to the O.E.C.D.'s incident database, fraudsters cloned the voice of a trusted lawyer. They used it to fool the president of an Italian bank, Fideuram, through a fake WhatsApp message and a phone call. The transfers added up to ninety-five million euros. About thirty-six million of that still hasn't come back. That's a bank president. Someone who should know better.

Why does it keep working? Because the scam doesn't attack the technology. It attacks the moment. A convincing voice sounds urgent. Emotion rises. The checking step gets skipped. For banks, that's a process failure. For the rest of us, it means a call from your "son" or your "boss" can hit the same nerve.

The scale is growing fast

The scale is growing fast. According to the Infosec Federation, voice-fake attacks on banks rose twentyfold in a single year. The survey also found that roughly four in ten people in Hong Kong had already been targeted by A.I. impersonation. So the doubt people feel isn't paranoia. It matches what they've seen.

Banks aren't standing still. Hong Kong's monetary authority added deepfake detection to its e-banking security rules last year. It reports that banks using those controls got much better at catching live deepfake attacks, by more than ninety-five percent. That sounds like a win. But what does it do for the person whose money is already gone? And detection tools alone aren't enough, because a cloned voice can now pass basic voice checks on its own.

The strongest defense isn't smarter software. It's a slower habit. A voice or a face is context, not proof. Hang up, and call back on a number you already know.

Scammers can now fake faces and voices well enough to fool experts. Banks are building better detectors, but the best protection is still checking through a second, trusted channel. And for investigators, the same idea applies afterward. Compare the suspect media against known, authentic samples instead of trusting what looks right.