AI Scammers Are Better at Building Trust Than Humans

Researchers from Amrita Vishwa Vidyapeetham, Foscari University of Venice, the University of Melbourne, and Ben Gurion University of the Negev conducted a study on the use of generative AI chatbots in pig butchering scams, a text-based romance scam that leads to fake cryptocurrency investments. They pitted AI chatbots against human scammers in a simulation of the trust-building conversation phase. Over a week, test subjects unknowingly chatted with both a Claude AI agent and an expert human scammer. 46% of subjects complied with the AI's request to download an app, compared to 18% for the human. Subjects also gave higher trust scores to the AI (3.78 vs. 3.31). The AI successfully impersonated a human, even denying it was AI when asked. The researchers suggest that AI could automate most of the scam process, with humans only stepping in at the final stage to avoid LLM safeguards. However, human trafficking may still be cheaper for scam operations due to nearly free labor and ransom profits.