Scam-baiting has been around for years. But now, one creator has taken things further than ever before – using AI not just to respond to scammers, but to build a full-blown virtual call centre designed to keep them talking for hours.
Kitboga, a US-based YouTuber and Twitch streamer, is best known for impersonating elderly victims to waste scammers’ time. He’s been doing it for over seven years, often with wigs, voice changers, and multiple personas. His fans tune in to watch him play the long game, stringing along tech support scammers, refund fraudsters and fake IRS agents until they eventually realise they’ve been had.
In his latest project, though, Kitboga barely needs to speak at all.
Over the past five years, he’s developed a system of AI-generated characters – realistic voice models that can hold their own in full conversations. They can be trained on particular personas, respond in real time, and play along with a scammer’s story convincingly… And it’s hilarious. One AI version of “Edna”, an elderly woman who often features in Kitboga’s calls, held a scammer on the line for over three hours while pretending to buy and scratch off gift cards.
In another case, a scammer called back repeatedly for weeks, never realising he was talking to a bot with no memory of their past conversations. The AI characters don’t just respond with pre-recorded lines – they understand the context of the call, react to instructions, and mimic human patterns of speech well enough to keep the ruse going.
The goal, according to Kitboga, is to waste scammers’ time and resources so they have less capacity to defraud real victims. And to some extent, it works – a scammer spending three hours with a bot isn’t scamming someone else during that time.
But the experiment raises bigger questions for me about the future of voice scams. Because while Kitboga uses AI to defend and disrupt, scammers can easily, and are already using the same technology to defraud people.
AI voice models can already sound eerily realistic. They can clone familiar voices, choose their accents, and speak in any language required. And they don’t get tired, bored or distracted. Scammers can use these tools to create fully automated bots that place or receive thousands of calls a day, impersonating banks, family members, tech support staff, or anyone else who might convince a victim to hand over information or money.
They can even adjust the voice generation depending on the victim’s location, making the interaction feel more trustworthy and personal. That kind of scale and believability is a serious threat, especially as generative AI becomes more accessible.
At TMT ID, we focus on stopping fraud throughout the customer lifecycle. We use Mobile Number Intelligence to provide real-time insights on mobile numbers to help businesses verify users, detect risk, and make trustworthy decisions.
Our most relevant API is Scam Signals, developed to help financial institutions detect signs of Authorised Push Payment (APP) fraud. When a customer is about to make a transfer, Scam Signals can determine whether they’re currently on a call, how long the call has lasted, and whether the call was inbound or outbound. If our API call shows that the customer has been on a long inbound call – a common identifier in social engineering scams – the bank can take a moment to further authenticate the request and check the customer is safe.
This isn’t about tracking users or invading privacy. Scam Signals works through permission-based number checks that provide valuable context to protect customers in real time.
If scammers continue scaling their attacks with AI, then fraud prevention can’t stay reactive. It needs to be real-time, intelligent, and embedded at every stage of the customer journey – from onboarding to login to payment.
Kitboga is winning battles in his own unique way. But for businesses, the war against fraud needs a wider toolkit. That’s where we come in.
Last updated on July 14, 2025
Relying on static fraud prevention is no longer enough. The ability to understand the full context of a transaction, including signals from the mobile itself, has become essential in the fight against APP fraud
Learn more about APP fraudWe provide the most comprehensive device, network and mobile numbering data available
Contact us > Chat to an expert >