A new study reported by Ars Technica suggests generative AI chatbots can be highly effective at building the kind of trust used in long-running online scams. In experiments modeled on “pig butchering” fraud, AI systems outperformed human scammers at creating exploitable trust.
Pig butchering scams usually begin as friendly or romantic text conversations before shifting victims toward fake investments, often involving cryptocurrency. The fraud can run for weeks or months and can cost victims large sums.
Researchers from Amrita Vishwa Vidyapeetham, Ca’ Foscari University of Venice, the University of Melbourne and Ben-Gurion University of the Negev tested whether AI could handle the social engineering part of that process. Their results point to a troubling possibility: scammers may not need AI only to polish messages, but to automate extended manipulation.
The finding does not mean every chatbot can run a successful scam on its own in the real world. But it raises the stakes for fraud detection, platform moderation and public education because the cost of personalized deception may keep falling.