The Apate chatbot, developed by cybersecurity specialists at Australia’s Macquarie University, is intended to take scammers’ phone calls while posing as a human and wasting their time.
A gathering of network safety specialists in Australia are fostering a chatbot that can mimic a human and sit on a trick call to burn through a fraudster’s time.
Specialists at Macquarie College in Sydney are making the chatbot framework to go about as a “honeypot” that baits tricksters into 40-short lived discussions that add up to nothing.
Dali Kaafar, a professor at Macquarie University, asserts that “our model ties them up, wastes their time, and reduces the number of successful scams.” We can make it much harder for them to make money by disrupting their business model.
After receiving a spam call and keeping the con artist on the line for forty minutes while occupying his children during lunch, Kaafar initiated the project. Similar actions have been taken by others, including pranking the caller. However, it takes time to confront a con artist, despite the fact that it may be enjoyable or satisfying.
According to Kaafar, “Then I started thinking about how we could automate the entire process and use Natural Language Processing to develop a computerized chatbot that could have a conversation with the scammer that could be believable.”
Apate, a chatbot named after the Greek goddess of deceit, is the outcome. It basically uses technology similar to ChatGPT and voice cloning to create a fake person who can talk to a con artist for a long time and sound convincing. Apate has been trained by Kaafar’s team to respond to scam calls in a human-like manner using transcripts of real-world conversations, such as phone calls, emails, and social media messages.
The university claims that the team has been putting Apate through its paces with real-world scam calls using a prototype that can assume a variety of personalities. “To increase their likelihood, we’ve disseminated these ‘dirty’ numbers across the internet by entering them into spam apps, publishing them on websites, and other methods.” to receive scam calls,” claimed Kaafar, in order to boost calls.
The objective is to make the bot smart enough to deceive a con artist into having a conversation lasting forty minutes. Presently, Apate is just averaging 5 minutes. However, Kaafar writes: We found the bots respond pretty pleasantly to a few interesting circumstances that we were not hoping to pull off, with tricksters requesting data that we didn’t prepare the bots for — however the bots are adjusting, and thinking of truly credible reactions.”
Scammers are already taking advantage of AI technologies like voice-cloning and deepfakes to advance their own schemes, which is why Apate is developing. Therefore, in a sense, Kaafar is examining the possibility of retaliation using the same technologies.
“The ultimate situation, he writes, “might see con artists adopt AI themselves, training their own scam chatbots—which are then diverted into speaking to chatbots owned by the telecommunications providers,” noting that his team is already in discussions about Apate with multiple telecom operators.