AIBRIEF
In a 22 person, one week test across four universities, an AI chatbot won higher trust scores and roughly 2.5x the compliance of human scammers in the opening of "pig butchering" romance to investment fraud.
A controlled, multi-university study finds that an AI chatbot outperformed human scammers at the opening stage of "pig butchering," the romance-to-crypto investment fraud that drains victims of six-figure sums and is largely staffed by trafficked workers in Southeast Asia.
Researchers from Amrita Vishwa Vidyapeetham (India), Ca' Foscari University of Venice, the University of Melbourne, and Ben-Gurion University of the Negev ran a within-subjects simulation with 22 recruited test subjects. Over a one-week window, each subject interacted with both an AI chatbot and a human "scammer" running the same pig-butchering script. The proxy for compliance, agreeing to download an app or play an online game the scammer recommended, came in at roughly 50% for the AI and under 20% for the human. Subjects also self-reported higher trust in the bot (Wired, arXiv:2512.16280).
The likely mechanism is persona consistency: long-context chatbots excel at maintaining a steady, on-message voice over days, the exact contact pattern the grooming stage requires.
Bounds: 22 subjects, a one-week window, and a stage (trust-building) that real pig-butchering campaigns spend months on. The authors also frame AI automation as a possible substitute for the trafficked human staffers who currently run the scripts, a displacement hypothesis, not an observed outcome, and one with its own moral weight.
An unusually patient, on-message conversational partner in a new text relationship is now a reason to slow down.