If you think you’d spot an A-I deepfake scam instantly, you’re not alone.
But new research by CommBank suggests Australians may be far more confident than correct, and scammers are thriving in that gap.
Because with deepfakes, the danger isn’t just the technology.
It’s what it hijacks - trust.
At Commonwealth Bank, General Manager of Group Fraud Management Services James Roberts says deepfakes are now being used to impersonate the kinds of people we’re wired to believe, whether that’s someone famous, someone familiar, or someone we love.
“So what we've been talking about in the study is deep fake. So like you said, either video deep fakes or audio deep fakes. So, often these are impersonations of someone you trust, so it might be a celebrity, a public figure, maybe even a family member. And they're often the contents often hosted on social media platforms, dating apps, and even messaging platforms.”
Mr Roberts says what makes this moment different is the scale, and the speed.
These aren’t clumsy scams you can laugh off.
They’re designed to feel personal, familiar, real.
And he says Australians are encountering deepfakes far more than many people realise.
“At the minute, according to the study, We're seeing over one in four Aussies, having seen a deep fake, and that could have been a video, deep fake or an audio, deep fake. So quite prevalent, but our expectations are that it should continue to grow as more and more criminals are also adopting this technology.”
One of the most unsettling aspects is how little material scammers need to create an audio deepfake - especially for people who already have voice recordings online.
“One of the things that is a challenge for all of us out there is that a lot of us are on social media, maybe we're in the public light or public view or we've posted online. So often our face and our voice print are out there. And it can only take sometimes, literally, under 10 seconds of someone's voice print. to copy their voice.”
So even if you’re “tech savvy”, Mr Roberts says the bigger risk is overconfidence - the assumption you’ll always know what’s real.
“One of the things that's worrying from the study is that 89 per cent of the respondents said that they could accurately spot a deep fake. But when we tested them… They were only correct 42 per cent of the times. That's actually worse than if they'd guessed.”
So if most of us can’t reliably detect a deepfake, what’s the best defence?
Mr Roberts says people don’t need to live in fear, and they don’t need to become experts in spotting synthetic voices or fake videos.
Instead, he says the best protection is behavioural - slowing down, checking, and refusing to be rushed.
“We don't need to be afraid. We actually don't need to be able to detect that it's a deep fake as long as anytime we're doing something like an investment or getting into a big relationship, that we take the simple stop-check-reject mantra steps. So, stop. Most scammers will put you under sort of time pressure to act without thinking. We've got a pause, take ourselves into slow-thinking mode, not react instantly. Second bit check. Talk to your friends and family before making a big financial decision. Use a search engine to look up the opportunity and add the word scam at the end. And then lastly, the reject portion… Hang up… block the calls.”
That advice matters because the stakes are enormous, not just financially, but emotionally.
Mr Roberts says scams can destroy relationships, break families apart, and in the most tragic cases, push people into self-harm.
“Yeah, there's the consequences are horrific and it's completely unconscionable that the syndicates would cause this amount of emotional harm. So obviously there's the financial harm, which is very significant, but what's often not talked about is the bits that you're mentioning there, the emotional harm. You've seen divorces, family breakups, We've seen small businesses that were emerging and doing well going under, and then in the rarer but still present side. There's, you know, self-harm incidents as well where people just feel they can't deal with it.”
He says that’s why talking openly about scams is so important, because shame and silence can make the damage worse.
“One of the best ways is to talk. Whether you've been scammed or you're just wanting to keep your family safe is to talk about it. We need to destigmatise scams so that people can have those open conversations that both help them not fall for it and if they do fall for it, make sure that they're truly cared for.”
And for families worried about voice-cloning scams, like someone pretending to be your child or grandchild asking for urgent money, Mr Roberts says one simple step can stop a scam in its tracks: a family safe word.
“So in fact, the studies showed that 74 per cent of Aussies felt that they should have a family safe word. But unfortunately only 20 per cent of them. Had actually taken the step to do that in such an easy step. So, If the listeners take one thing out of the whole discussion here, make a safe word up, share it confidentially within your family to help stay safe against any impersonation of any of your family members.”
But Mr Roberts also has practical advice for the everyday moment - the phone rings, it’s an unknown number, and you’re tempted to answer just to see what they want.
He says don’t engage, don't even take the opportunity to tell the scammer off.
“So so while it probably is satisfying to cuss them out that and have that engagement. It's often best not to answer a scam call if it's coming in. The more you answer it, they feel that maybe there's an opportunity for scamming you, but if you're never answering it, I think practically you end up dropping off their lists and eventually they stop calling you.”
Now, the story doesn’t end with scammers using A-I to scale up their deception.
Because the same technology is also being used to fight back.
At Commonwealth Bank, Mr Roberts says they’re using generative AI to build systems that don’t just detect scams, but actively interfere with them.
He says they’ve partnered with a specialist group to create bots that engage scammers directly, keeping them busy, draining their time, and gathering intelligence that can be used to protect real people.
“One of the more exciting things is that obviously I've talked a little bit about the bad things that generative AI can be used for, but we're also, at the CommBank, we're giving it a go of using generative AI for good to try and stop the scams too. So we've partnered up with a company called Apate AI… And we've created our own little bot army of voice bots and messaging bots that actually go out and engage with the scammers, wasting their time, so every minute they're talking to a bot, they're not talking to an Aussie and defrauding them.”
Mr Roberts says the bot conversations can also extract key details, like dodgy bank accounts or links, which can then be blocked, potentially stopping multiple victims.
“And we also use it just like you were saying now, we scrape intelligence too. So, you know, if the bot's told a bad bank account to deposit into or take that intelligence and we use it to block. So one piece of intelligence could save maybe 10 other Aussies.”
And sometimes, the bot doesn’t just keep a scammer busy for a few minutes, it can hold them there far longer.
“Our record at the minute is, 54 minutes of holding the person on the line for, which is, yeah, it's 54 minutes they can't be getting at another Aussie.”
Cyber and scam support specialist Simon Smith from Scam Assist, a company that helps in the recovery of funds lost to online scams, fraud, and financial cybercrime.
Mr Smith says the double-edged nature of A-I - can be used to amplify fraud, but it can also be used to disrupt it.
Mr Smith says people are already benefiting from automated detection in simple ways, like phones flagging suspicious numbers, even if they don’t think of it as A-I.
“You know, if you've got an Android phone, your phone automatically, marks calls as suspicious or, if you want to get technical, I mean, that's technically just an algorithm that's saying, this number is known for fraud, and it's accurate to 99 per cent of the time.”
But Mr Smith also warns people not to get hypnotised by the buzzword.
Because scams aren’t new - only the packaging is.
“AI is a very broad term, is a valid one, right? And where there are really good. There's really good technology out there. And there's bad scammers, and there's bad technology and good scammers. And it doesn't matter, whether there's AI or not, there'll always be fraudsters, and, you know, as long as there's greed, there'll be theft."
Deepfakes may be the latest weapon, but the best protection is still deeply human:
Slow down.
Check with someone you trust.
And don’t let urgency override your instincts.
