Artificial intelligence is no longer just a tool for cybercriminals. It is now being deployed to fight back against digital fraud by wasting the time and resources of scammers. Australian company Apate has built a system that diverts phone scammers onto calls with AI bots trained to keep conversations going as long as possible. The goal is to give scammers enough hope that they stay on the line, while simultaneously collecting intelligence about the scams they are running.
Apate AI Bots Trap Scammers
Dali Kaafar, the founder and CEO of Apate, describes the strategy as building the perfect victims for scammers. He notes that every minute a scammer spends talking to a bot is a minute they are not reaching out to real people. The platform currently operates around 350,000 bots. These agents do not just answer phone calls. They also infiltrate scam chat groups online and respond to text messages with similar goals of frustrating scammers while gathering data.
The system has collected more than 250,000 pieces of information about fraudsters in real time. This data includes scam URLs, money mule accounts, and bank details. The bots are designed with different personalities, language skills, and profiles to avoid detection. They mimic human behavior by sometimes having WhatsApp, sometimes picking up the phone, and sometimes hanging up to say they will call back later.
Testing of the tool reveals that the AI is convincing enough to frustrate experienced users. In a demo where testers played the role of scammers, the AI maintained a natural conversation flow. The bots expressed skepticism but left openings for the scammers to continue their pitch. The testers could not get the AI to invest in a cryptocurrency scheme, even after six minutes of effort. The conversations often last more than two hours, according to Kaafar.
Scam Intelligence and Defense
Cybercriminals initiate billions of messages and calls each year. Professional scambaiters and law enforcement have struggled to stop the expansion of digital scamming. AI offers a new way to disrupt these operations. By exploiting the psychological vulnerabilities of criminals, defenders can waste their resources and prevent them from launching other attacks.
This approach extends beyond phone calls. Security researchers are using large language models to power honeypots. These are false virtual machines designed to attract hackers. Mark Vero, a doctoral researcher at ETH Zurich, found that LLM-powered honeypots keep AI agents attacking the system significantly longer. The attackers are more convinced by the simulated environment and mark them as fake at a lower rate.
Better intelligence sharing between police, social media companies, and banks is desperately needed. AI-driven monitoring and data analysis could aid this effort. The use of generative AI to keep criminals preoccupied appears to be an increasingly effective strategy in the fight against online crime.
Source: Wired
