Governments worldwide have struggled to curb online crime that easily crosses borders, and the rise of artificial intelligence has simultaneously amplified scam operations while offering new defensive tools. This tension has inspired experiments that flood scammers with automated responses, hoping to waste their resources and reduce the number of real victims reached.
Apate, named after the Greek goddess of deception, has spent the past two years building a platform that redirects phone scammers to AI-driven agents designed to sustain conversation without ever conceding to the fraud. The goal is to give scammers enough hope to stay on the call while the system quietly records details about the scheme, turning each interaction into a source of actionable intelligence.
The service now runs roughly 350,000 bots, a figure supplied by the company’s chief executive, and is integrated with banks and telecom operators. In real time the bots have harvested over 250,000 data points, ranging from malicious URLs to mule accounts and compromised bank details.
Each bot is given a distinct persona, language capability and online profile so that scammers are less likely to spot the artificial nature of the interaction. Some bots appear on WhatsApp, others do not; a few will answer a call and later claim they will return, mimicking human hesitation.
A test by the security outlet Kernel Panic let volunteers act as scammers against a demo version of the system, revealing that the bots can keep the dialogue going while frustrating the attacker. Cybercriminals still generate billions of scam calls and messages each year, and traditional law-enforcement and research initiatives have not halted the overall growth of digital scamming.
Researchers suggest that AI could also improve cross-industry intelligence sharing among police, banks and social platforms, while open-source honeypot projects are now embedding large language models to make fake machines more convincing, according to ETH Zurich doctoral researcher Mark Vero. Such generative-AI honeypots aim to waste attacker time, gather tactics and ultimately make cybercriminal operations less efficient.