Apate Deploys 350,000 AI Bots to Waste Scammers' Time

Australian security firm Apate has built a defense platform operating roughly 350,000 specialized AI bots designed to intercept phone calls, text messages, and online scam groups. Backed by telecommunication providers and utilized by major banking institutions, the system impersonates prospective victims to keep fraudsters engaged in conversations that frequently extend past two hours. By keeping bad actors occupied, the automated platform prevents them from reaching real targets while gathering intelligence.

Apate assigns varied personalities, language profiles, and online behaviors to its bots, allowing them to display realistic skepticism without entirely discouraging the fraudster. Through these prolonged automated interactions, the platform has successfully harvested more than 250,000 pieces of real-time intelligence on active criminals, including scam URLs, money mule accounts, and specific banking details.

This defensive paradigm is also expanding into system security. Research led by Mark Vero, a doctoral researcher in the department of computer science at ETH Zurich, reveals that integration of large language models into open-source honeypots fundamentally alters hacker interactions. Vero and his team discovered that LLM-powered honeypots kept AI attack agents engaged significantly longer and were flagged as decoy systems at a much lower rate compared to standard, predictable virtual setups.

For security operations and threat intelligence teams, these generative AI techniques present a pragmatic shift toward active resource exhaustion. By deploying responsive, context-aware agents across consumer channels and network decoys, defenders can drain criminal infrastructure, harvest high-fidelity operational data, and disrupt automated attack pipelines before they cause harm.