Deepfake: 1 in 15 Australians Hit, and Familiar Voices Lie

One in fifteen Australian adults says someone has made an A.I. fake image of them since they turned eighteen. That's a new finding from researchers at R.M.I.T. University. And nearly nine in ten of those victims say it hurt their health, their work, their relationships, or their money.

If you've ever posted a photo, or sent a voice

If you've ever posted a photo, or sent a voice message, this story is about you. It's unsettling, and it's fair to feel that way. The R.M.I.T. study looks at what's called image-based abuse, meaning fake or stolen pictures used to hurt someone. Researchers found that most victims of sexually explicit deepfakes are women, so this isn't random. Meanwhile, the same tools are now being used to steal money. So if a face or a voice can be faked this well, what can we still trust?

Start with the number itself. One in fifteen sounds small until you picture a full train carriage. In a crowd that size, a few people have already lived this. And for most of them, the damage didn't stay online. It followed them into work, family, and bank accounts.

Now the money side. According to a report tied to Australia's financial regulator, scams using a deepfake of the Prime Minister have cost victims about seven point four million dollars. The fake leader pitched big investment returns. People saw a familiar face, heard a familiar voice, and sent money. That's the whole trick. Familiar feels safe.

Can people just spot the fakes

So can people just spot the fakes? Not reliably. A review of nineteen studies, covering about twenty-four thousand people, found that roughly one in four can't tell a Hong Kong Police Face Scans Start No Word On Wrong Matches from the real thing. It also found people wrongly call real video fake nearly as often. So the problem runs both ways. Real footage gets doubted, and fake footage gets believed.

Maybe software can do better. Not yet, at least for voices. One roundup of 2026 statistics puts some voice detection tools as low as thirty-eight percent accurate. That's worse than flipping a coin. And fakes that pair a cloned voice with video can slip past standard visual checks. For the rest of us, that means a call that sounds exactly like your daughter, or your bank, proves nothing on its own.

For investigators, it changes the job. A face match that looks perfect can fall apart in court. A defense lawyer only has to argue the original image might be A.I. So the first question is no longer whether two faces match. It's where the picture came from. For everyone else, that's the same question to ask about the clip in your feed.