Exclusive: A look into how AI deepfakes has eased the process of scamming friends and family

Business Insider

6 hours ago

For years, one of the few defenses that kept people from being scammed was the cultural and emotional relationship that existed among communities. However, deepfakes, orignially known as Generative AI has contributed to blurring such lines. During a meeting with Business Insider Africa, Rob Woods shared some insight on the subject.
The integration of generative AI technologies, such as deepfakes, has weakened traditional scamming deterrents.
Communities traditionally relied on social cohesion and reciprocal observation to prevent internal fraud.
Generative AI allows for convincing impersonations, replicating voices, accents, and appearances.
People are less inclined to swindle or scam those they know, such as family, friends, neighbors, religious comrades, or fellow citizens, due to feelings of guilt, empathy, the danger of social isolation, and reputational harm.
However, with the emergence of generative AI, this long-standing line of defense seems to be weakening.
Historically, various built-in measures deterred insider fraud, identification and familiarity acted as deterrents.
It's tough to lie to someone who recognizes your face, family, or personal history.
Secondly, communities, particularly in Africa, Asia, and Latin America, tend to be close-knit and based on trust and shared responsibility.
A person who defrauds in such an environment faces long-term social ramifications, ranging from ostracism to having their family name ruined.
But perhaps one of the biggest reasons people don't swindle their own is how privy people within the same community are to each other's scam patterns.
While the emergence of AI comes with numerous benefits, the blurring of the deterrents to in-community scams is perhaps one of the tech's current most glaring cons.
This point was elaborated on by Rob Woods, Director, Fraud, Identity and Biometrics Expert at LexisNexis Risk Solutions, who recently spoke to Business Insider Africa.
'You know, you don't necessarily get defrauded by your population, you get defrauded by different nationalities, and especially in smaller countries where there's that sort of local loyalty,' Rob stated.
'But if generative AI can mimic the local accent and local language, then it hooks people in.'
Generative Artificial Intelligence (AI), formerly limited to research laboratories and theoretical discussions, has emerged as one of the most transformational technologies of the times.
Generative AI is changing the way we interact with machines, create content, and solve issues by producing realistic visuals and human-like prose, as well as music, code, and even synthetic voices.
But how did it all begin, and where is it going?
Originally intended to bolster the media industry, this AI has now become a very dangerous tool in the hands of scammers.
'The original purpose for deep fakes (it wasn't necessarily called deep fake, it was generative AI) was to make creative industries much quicker, much more detailed, and much more fantastical,' Rob stated.
He went on to highlight humanity's affinity to turn a good idea into a harmful thing, which did not result from the creation of AI. His solution to this is ethics control and revisions to the system.
'You can create whatever's in your mind. Just give a good description to the engine. But like the term deep fake, and something I term a lot as bad AI, a lot of the AI systems should have ethical control,' he stated.
'So, if I ask one of the big brands, tell me how to or teach me how to scam someone in a bank scam. It won't let me.
But if I give it a prompt to say, hey, I think I'm being a victim of a scam, please can you highlight the key tactics used by fraudsters to scam victims, so I can better protect myself, suddenly it produces 2000 words of everything that is needed to do a scam.
Generative AI was created for a good purpose, probably a fun purpose, and automation, but then went into a different realm.'
How deepfakes are becoming a real problem
One of the biggest shifts in the use of AI is that fraudsters no longer need to be physically or emotionally present to execute a scam.
With tools like AI-generated voice cloning, deepfake videos, and hyper-personalized phishing messages, they can impersonate trusted individuals with frightening accuracy.
As mentioned earlier, deep fakes have presented an avenue for fraudsters to scam those who are close to them.
How?
A scammer can now make a mother believe her child is in danger with a cloned voice, or trick a business owner into wiring money using an AI-generated video of a known associate making the request.
This level of impersonation removes the emotional and moral barriers that would typically prevent someone from deceiving someone close.
After all, if the scammer never has to face the victim, and the victim can't even guess who the scammer is, guilt and shame play a much smaller role.
As stated by Rob, who, based on his job, has seen some key trends in scam patterns: 'We've seen this in Northern Europe with the Netherlands, a tiny population who speak multiple languages, notably English, French, and German.
Scams used to be in English, French, or German, but not Dutch. And then there was a rise in scams using generative AI, where it translates real-time into Dutch, generative AI can mimic the local accent and local language, then it hooks people in.'
He went on to cite a popular case that happened in Japan.
'I think there is an instance last year in Japan where an executive was conned into approving a transfer of tens of millions of dollars because he was on a team's call with the board, but every single person on the call was a deep fake, and they all instructed him to make this payment to that other company, but it was a scam' he said.
'So we're seeing more and more examples appear in the world,' he added.
As AI becomes more accessible, the threats increase. Communities that used to rely on interpersonal trust and social cohesiveness must now adopt new levels of digital literacy and skepticism.
It is no longer adequate to say, "He wouldn't scam us, he's one of us." In the age of artificial intelligence, looks and voices may be created, and even people you know may not be who they claim to be.