AI scammers outperform humans when it comes to building trust
The AI chatbot was more effective at creating “exploitable trust” than the humans.
A new study from four universities suggests that generative AI chatbots can outperform human operators in the trust-building phase of "pig butchering" scams, the long-running romance-to-crypto cons that drain six-figure sums from victims worldwide.
The researchers, drawn from Amrita Vishwa Vidyapeetham, Foscari University of Venice, the University of Melbourne, and Ben Gurion University of the Negev, ran a controlled simulation pitting AI chatbots against human scammers. The bots handled the relationship-establishment stage, which in real operations can stretch for months, and by several measures built "exploitable trust" more effectively than their flesh-and-blood counterparts.
The implication is structural. Fraud operations have always been labor-intensive: recruiting, training, and retaining social engineers willing to spend hours in chat windows pretending to be someone they are not. AI collapses that labor cost. A single model can run thousands of parallel conversations, maintain consistent personas, and never tire, complain, or quit.
For the security industry, this shifts the defensive calculus. Detection tools trained on human typing patterns, response latency, or emotional cadence may find those signals evaporating as bots produce increasingly natural dialogue. The bottleneck in pig-butchering operations moves away from human recruitment and toward victim acquisition, the one stage where a convincing online presence still matters.
The study also reframes a familiar debate. The question is no longer whether AI will assist scammers, but whether the human scammer is becoming optional. If the trust-building phase, historically the hardest part to scale, is now the part machines handle best, the economics of organized fraud tilt sharply in favor of those who deploy them.