Anti-cybercrime initiatives are increasingly deploying artificial intelligence to disrupt scammers by engaging them with lifelike bots, which the criminals mistake for genuine victims. This strategy aims to waste scammers' time and resources while gathering intelligence on their operations.
For the past two years, the Australian company Apate, named after the Greek goddess of deception, has been developing a system that diverts phone scammers into extended conversations with AI bots. These bots are designed to maintain engagement without ever falling for the scam, giving scammers false hope and keeping them on the line for as long as possible.
Dali Kaafar, Apate's founder and CEO, stated that the platform effectively creates "perfect victims" for scammers. He explained that every minute a scammer spends interacting with a bot is a minute saved for potentially hundreds or thousands of real individuals who might otherwise be targeted. Apate's platform, which is utilized by banks and supported by telecom companies, reportedly employs around 350,000 bots.
These bots not only answer calls but also infiltrate online scam chat groups and respond to text messages, with the dual objectives of frustrating scammers and collecting intelligence. Kaafar claims the company has amassed over 250,000 pieces of real-time information about fraudsters, including scam URLs, money mule accounts, and bank details.
To enhance their realism, the bots are programmed with diverse personalities, language skills, and profiles. They exhibit varied behaviors, such as having WhatsApp, picking up the phone, or even hanging up and promising to call back later, mimicking human unpredictability. Apate's system has demonstrated its ability to sustain calls for over two hours.
While AI is increasingly being used to supercharge digital scams, it is also being recruited to protect potential victims. Governments worldwide have struggled to combat cross-border online crime, leading to alternative approaches like using automation to target scammers en masse, for instance, by spamming spammers to deplete their resources.
Beyond Apate's efforts, the broader trend includes the use of generative AI to preoccupy cybercriminals, thereby wasting their resources and preventing them from launching other attacks. This approach is reminiscent of honeypots, which companies and security researchers have long deployed as false virtual machines to attract hackers, gather information, and understand their techniques.
Mark Vero, a doctoral researcher at ETH Zurich's department of computer science, noted that open-source honeypot providers are increasingly integrating large language models (LLMs) into their systems to make them more realistic. Recent research by Vero and his colleagues found that an LLM-powered honeypot could keep AI agents attacking their system "significantly longer" than a honeypot with more predictable behavior. The study also indicated that agentic attackers were more convinced by LLM-simulated honeypots and identified them as actual honeypots at a much lower rate.
Despite these advancements, cybercriminals continue to initiate billions of messages and calls annually, with sophisticated operations often running industrial-scale scamming sites. While professional scambaiters and initiatives to infiltrate scam operations have provided critical deterrence for years, law enforcement and research efforts have not halted the overall expansion of digital scamming. However, AI could usher in a new era for anti-scam efforts, potentially aiding intelligence sharing between law enforcement, social media companies, banks, and other industries through AI-driven monitoring and data analysis.
