A scammer calls and asks for a password. Malcolm, an older man with an English accent, is confused.
—What business are you talking about? —Malcolm asks.
Another day, another scam phone call.
This time, Ibrahim, a cooperative and polite man with an Egyptian accent, answers. “Frankly, I’m not sure I remember buying anything recently,” he tells the hopeful scammer. “Maybe one of the guys did,” Ibrahim continues, “but it’s not your fault, is it?”
Scammers are real, but Malcolm and Ibrahim are not. They are just two of the conversational AI bots created by Professor Dali Kaafar and his team. Through his research at Macquarie University, Kaafar founded Apate, named after the Greek goddess of deception.
Apate aims to defeat global phone scams with conversational AI, leveraging existing systems where telecom companies divert calls they can identify as coming from scammers.
Kafaar was inspired to turn the tables on phone scammers after playing a “dad joke” on a scammer who called in front of his two children while they were enjoying a picnic in the sun. With silly small talk, he kept the scammer on the line. “The kids laughed a lot,” he says. “And I thought the purpose was to trick the scammer, to waste his time so he wouldn’t talk to others.
“Swindling the scammers, so to speak.”
The next day, he called his team at the university’s Cyber Security Center. He figured there had to be a better way than his “dad joke” method. And there had to be something smarter than a popular piece of existing technology: the Lennybot.
Before Malcolm and Ibrahim, there was Lenny.
Lenny is an older, lazy-looking Australian with a desire to chat non-stop. He is a chatbot designed to troll telemarketers.
In a raspy voice, punctuated by a slight hiss, Lenny repeats several phrases on a loop. Each phrase begins after a second and a half of silence, to mimic the rhythm of a conversation.
Lenny’s anonymous creator posted on Reddit who created the chatbot to be “a telemarketer’s worst nightmare… a lonely old man who is willing to chat, proud of his family, and can’t focus on the telemarketer’s goal.” The act of roped in scammers has been dubbed scambaiting.
Apate robots to the rescue
Telecoms companies in Australia have blocked almost 2 billion fraudulent phone calls since December 2020.
Thanks in part to $720,000 in funding from the Office of National Intelligence, there are now potentially hundreds of thousands of “victim chatbots” — too many to name individually. The bots of various “ages” speak English with a variety of accents. They have a range of emotions, personalities and responses. Sometimes they are naive, sometimes skeptical, sometimes rude.
If a telecom company spots a scammer and diverts them to a system like Apate, the bots will work to keep the scammers busy. They try different strategies and learn what works to ensure the scammers stay on the line longer. Through success and failure, the machines hone their pattern.
As they do this, they extract intelligence and detect new scams, gathering information on how long the call lasts, when scammers are most likely to call, what information they are looking for, and what tactics they are using.
Kafaar hopes Apate will put an end to the scam calling business model, which is often run by large, multi-million dollar criminal organisations. The next step is to use the information obtained to detect and address scams in real time.
“We are talking about real criminals who make our lives miserable,” says Kafaar. “We are talking about the risks that real human beings are taking.
“Human beings who sometimes lose their life savings, who can be paralyzed by debt and sometimes psychologically wounded by shame.”
Richard Buckland, a professor of cybercrime at the University of New South Wales, says technology like Apate is different from other types of scams, which can be amateurish or amount to vigilantism.
“Normally, scambaiting is problematic,” he says, “but this is clever.”
After the newsletter promotion
Mistakes can be made when people take things into their own hands, he says.
“You can target the wrong person,” he said. Many scams are carried out by people in conditions of servitude, almost slavery, “and they are not the ones on the wrong side.”
“Some scammers are tempted to go further and take matters into their own hands, to fight back or to confront them. That is problematic.”
But, he says, the Apate model appears to be using AI for good: as a kind of “trap” to lure criminals in and then learn from them.
Buckland warns that a high level of confidence would be needed that telecoms companies are just diverting fraudsters to AI bots, because misidentification happens everywhere. He also warns that criminal organisations could use anti-scam AI technology to train their own systems.
“The same technology used to fool scammers could be used to fool people,” he says.
The National Anti-Scams Centre (NASC) runs Scamwatch under the auspices of the Australian Competition and Consumer Commission (ACCC). An ACCC spokesperson says scammers often impersonate well-known organisations and can often spoof legitimate phone numbers.
“Criminals create a sense of urgency to try to get victims to act quickly,” the spokesperson said. “They often try to convince victims to share personal or bank account details, or to give them remote access to their computers.
“Criminals may already have some details about their intended victims, such as their name or address, which they obtained or purchased illegally through a data breach, phishing or other scam.”
This week, Scamwatch had to issue a warning about a kind of meta-scam.
Scammers claiming to be from NASC itself would call innocent people and tell them they were being investigated for being involved in a scam.
NASC says people should immediately hang up the phone against scammers and “not attempt to engage with criminals.” The spokesperson said he was aware of “technological initiatives to produce scam baits using AI voice characters,” including Apate, and would be interested in reviewing any evaluation of the platform.
Meanwhile, there is a thriving online community of scammers and Lenny remains one of their cult heroes.
In one memorable recording, Lenny asks the caller to hold on for a minute, as ducks begin quacking in the background. “I’m sorry,” Lenny says. “What were you saying?”
“Are you at your computer?” the interlocutor asks impatiently. “Do you have a computer? Can you come over to it now?”
Lenny continues until the scammer loses control: “Shut up, shut up, shut up.”
“Could you wait a minute?” Lenny asks as the ducks start quacking again.