Cybersecurity experts at Australia’s Macquarie University have developed a conversational artificial intelligence (AI) bot that can pretend to be a victim and keep scammers on the phone for a long time.
The chatbot, named Apate after the Greek goddess of deception, is multilingual and can be trained to adopt different accents and languages. It is now being tested on live scam calls.
Trained with recordings of scam calls, scam emails, chats, and other conversations, Apate uses voice cloning technology to play a convincing victim.
“We found that the bots react pretty nicely to some tricky situations that we were not expecting to get away with, with scammers asking for information that we didn’t train the bots for — but the bots are adapting, and coming up with very believable responses,” Professor Dali Kaafar, the executive director of Macquarie University’s Cyber Security Hub, said.
According to Professor Kaafar, the goal is for the chatbot to waste scammers’ time and protect potential victims. Currently, Apate can keep scammers on the phone for an average of five minutes, but the team aims to keep them for up to 40 minutes.
The Origin of Apate
Professor Kaafar came up with the idea to create a chatbot to “scam the scammers” after receiving a scam call while having lunch with his family and successfully keeping the scammer on the phone for 40 minutes.
“I realized that, while I had wasted the scammer’s time so they couldn’t get to vulnerable people, which was the point — that was also 40 minutes of my life I wouldn’t get back,” he said.
“Then I started thinking about how we could automate the whole process, and use Natural Language Processing to develop a computerised chatbot that could have a believable conversation with a scammer,” Professor Kaafar added.
Kaafar and his team analyzed the social engineering techniques scammers employ and used machine learning techniques and NLP (Natural Language Processing) to identify typical scam scripts and train the chatbot.
“The conversational AI bots we have developed can fool scammers into thinking they are talking to viable scam victims, so they spend time attempting to scam the bots,” Professor Kaafar said.
Apate is undergoing live trials. The researchers are redirecting scam calls to the bot.
“We’ve put these ‘dirty’ numbers all around the internet, getting them into some spam apps, or publishing them on webpages and so on, to make them more likely to receive spam calls,” Professor Kaafar revealed.
Fighting Scam Calls
Professor Kaafar said his team is open to commercial partnerships that could see Apate deployed widely.
“The scam-fighting bots also contribute to threat intelligence – timely information that is gathered about current phone scams and their targets; this helps organizations such as major banks, retailers, and government bodies warn customers,” the report said.
While phone scams have been around for many years, they’re not showing signs of abating. According to the Federal Trade Commission (FTC), phone scams yielded the highest loss per person (about $1,400 on average) in 2022.
Help desk fraud is one of the most common types of scam calls. It usually involves scammers, who pretend to be part of the tech support team of reputable companies, asking targets for sensitive information like their login details. Read our explanatory piece on help desk fraud to learn more about this scam and how to protect yourself.
