AI chatbots prove more persuasive than human scammers at building victim trust

Artificial intelligence is entering new territory in the world of fraud: a recent academic study demonstrates that AI chatbots can execute the trust-building phase of elaborate scams more successfully than human perpetrators. The findings raise serious concerns about the future weaponization of generative AI in schemes that already cost victims tens of billions annually.

The Experiment#

Researchers from Amrita Vishwa Vidyapeetham (India), Foscari University of Venice, the University of Melbourne, and Ben Gurion University of the Negev conducted a comparative test of AI-driven versus human-operated scamming tactics. They focused on pig butchering—a term for text-based romance fraud that eventually pivots toward fake cryptocurrency investment schemes, often extracting six-figure losses from individual targets.

The study directly pitted AI chatbots against human scammers in a controlled simulation, measuring their ability to establish rapport and persuade victims during the months-long preliminary relationship phase that typically precedes a financial solicitation.

The Results#

Over the course of a week, 22 unwitting test subjects engaged in conversations with both chatbots and human scammers. When asked to perform a proxy action—downloading an app or playing an online game—nearly half responded affirmatively to the AI bot, compared to fewer than 1 in 5 who complied with human operators. Participants also rated their trust levels significantly higher for the AI counterparts.

Why It Matters#

The findings underscore how generative AI can now autonomously manage the complex, extended relationship-building work that makes pig butchering scams so effective. Where AI was previously viewed as a tool to refine language or enhance deception, this research suggests it may fully replace—and outperform—human fraudsters in executing entire scam sequences. The implications for fraud prevention and victim vulnerability are profound.