How to avoid being scammed by artificial intelligence – Diana Clement

Online scams are getting worse by the week. Photo / 123RF

Warning: online scams are getting worse by the week. Artificial intelligence (AI) is being used to steal more and more money from Kiwis, and the technology is only set to get far more deceptive, making it harder than ever to distinguish fact from fiction.

Off-the-shelf software is already being used to create highly realistic video and audio fakes, aka deepfakes. They’re one new frontier in scams.

You get a call from a family member in trouble, and the video or voice looks and sounds real. What parent or grandparent wouldn’t want to help their child in an emergency? Or think of the staff member who believes the boss has just authorised a payment.

Likewise, automated robocalls are becoming more difficult to distinguish from real calls. Many people will genuinely think they’ve been talking to a human being who is responding to the conversation, rather than a bot.

Increasingly AI-driven cybercrime networks can operate at an unprecedented scale, running thousands of simultaneous scams with minimal human intervention. AI can correct the scammer’s English in real time. Chatbots can reply instantly and apparently intelligently, making you think you’re talking to a human.

People will find themselves being blackmailed with fake videos. How would a partner, or employer know that the compromising video of you was real or AI-generated? Blackmailers already know to keep their financial demands realistic so the victim will just pay, to avoid the pain of refusing, said Sean Lyons, chief online safety officer at Netsafe.

You can’t even scan a QR code now, without the risk of being sent to a phishing website where your financial information is stolen. Overseas, scammers have been sticking fake QR codes over genuine ones in parking lots. This could happen anywhere QR codes are used. Not just for car parking.

It’s not easy to avoid these scams. How do you park, if you’re too nervous to scan a QR code? Maybe it’s the only option in that location? How do you really know the 0800 number for your bank that you’ve just Googled is the actual bank number? Or that the person who has just called with the number withheld is actually from the bank?

“Bank impersonation scams are very convincing,” said Nicola Sladden, New Zealand’s Banking Ombudsman. In one example the Banking Ombudsman Scheme heard, scammers had spoofed the bank’s phone number, so the call appeared genuine. The bank declined to refund the money. In that case, however, the Ombudsman recommended the bank refund a $30,000 loss to the customer.

AI-driven bots can bypass traditional security measures like CAPTCHA. Scammers can create realistic copycat online shopping sites with ease. These sites have .co.nz web addresses, but are scams. Victims see an item for sale they’d like to buy, not knowing the site is either taking their money without sending the item or harvesting their card details for further fraud.

The technology can already generate virtual influencers that build trust over time before directing their followers into financial traps. They can even generate deep histories, making them appear legitimate.

Brand new, emerging scams are sometimes the worst because warnings don’t yet exist. The QR code scams are an example. Who would have thought twice when scanning a QR code until very recently?

Existing scams are metamorphosing, said Lyons. For example, romance scammers used to ask for money for their “operation” or other “emergency”. Now they suggest their new “partner” invests in crypto with them, which turns out to be a scam. Or they may use one of many older-style scams on the smitten victim.

Lyons points out that previous red flags such as poor English are things of the past.

“Technology is giving the scammers a chance to iron out all of those things that we might have traditionally seen as being red flags. So it’s removed the poor construction. It’s removed the bad logos. It’s removed the implausibility.” All of this can be generated by AI with ease.

While the efficacy of traditional red flags has been reduced, a combination of tried and tested ways to protect yourself can help reduce the risks, Lyons said.

Verify identities when receiving unexpected calls, emails or messages, even from someone you know. Stay informed with the latest scam tactics and AI advancements. Use multi-factor authentication whenever possible. Beware of urgent requests, and make sure all your devices have up-to-date security.