Cybersecurity researchers have unleashed an army of 200,000 artificially intelligent "victims" designed to engage online fraudsters in endless, time-wasting conversations, effectively turning the tables on scammers who prey on vulnerable people.
Fighting Fraud With Fake Targets
The project, known as Apate, deploys AI-powered chatbots that pose as gullible potential victims. These bots are programmed to engage scammers in drawn-out exchanges, keeping fraudsters occupied and diverting their attention away from real people who might otherwise fall for their schemes.
The scale of the operation is significant, with hundreds of thousands of these synthetic personas reportedly interacting with a massive number of scammers each month. By flooding fraud operations with fake targets, the initiative aims to make scamming less profitable and far more frustrating for those running the schemes.
The goal isn't just to waste scammers' time — it's to make fraud so infuriating that criminals give up.
Measuring Success in Frustration
Perhaps the most unusual metric behind the project is how its creators gauge effectiveness. Rather than relying solely on conventional performance indicators, the team tracks how often scammers lose their temper and resort to profanity when dealing with the persistent, maddeningly clueless bots.
This unconventional key performance indicator reflects a broader strategy: the more frustrated a scammer becomes, the more time and energy they've wasted on a target that will never pay out. The bots' ability to convincingly string along fraudsters speaks to how sophisticated conversational AI has become.
The approach represents a growing trend in the fight against online fraud, where defenders are increasingly using automation and AI to fight back against criminals who themselves have adopted more advanced tools.
- Roughly 200,000 AI "victims" are deployed against scammers
- The bots reportedly engage hundreds of thousands of fraudsters monthly
- Success is partly measured by how often scammers curse at the bots