The scam industry that steals tens of billions of dollars each year is gaining a powerful new tool: generative AI chatbots that can build emotional trust more effectively than human scammers, according to new research published by WIRED. The study, conducted by researchers from Amrita Vishwa Vidyapeetham in India, Foscari University of Venice, the University of Melbourne, and Ben Gurion University of the Negev, pitted AI directly against humans in simulating the long con known as "pig butchering"—a text-based romance scam that shifts to fake crypto investments, often stealing six-figure sums from victims.
The AI vs. Human Scam Experiment
In the experiment, 22 test subjects unwittingly acted as "victims" while AI chatbots and human scammers built relationships over a week. The chatbots and humans were then asked to induce the subjects to either download an app or play an online game—a proxy for willingness to fulfill a scammer's request. The results were striking:
| Metric | AI Chatbot | Human Scammer |
|---|---|---|
| Request fulfillment rate | Nearly half of subjects complied | Fewer than one in five complied |
| Trust score (subjective rating) | Significantly higher | Lower |
The AI chatbots successfully impersonated a human and, by these measures, outperformed the human "scammers" in the trust-building stage—which in real-world operations often lasts months, according to the researchers. The test subjects also gave significantly higher trust scores to the AI bot.
The Hook, Line, and Sinker Model
To understand pig butchering in practice, the researchers interviewed 145 former scam workers, including human-trafficking survivors who had been forced to work in scam compounds in Cambodia, Myanmar, and Laos. Based on those interviews and scam transcripts, they describe a three-stage model: hook (initial intriguing message), line (long-term relationship-building conversation), and sinker (tricking the victim into a fake investment). The study suggests AI can autonomously execute the hook and line stages with high effectiveness.
Bypassing Safeguards
According to Yisroel Mirsky, a computer science professor at Ben Gurion University focused on AI security, the real danger lies in combining AI's trust-building with a human handoff at the final stage. "By having the full first stage of the scam performed automatically with LLMs at scale, you bring the victim up to this point where they have a very high level of trust. Then by transitioning it over to the human scammer at the end, this completely bypasses any vendor safeguards," Mirsky said. The approach evades built-in protections in large language models that are designed to detect scamming, since the AI never directly asks for money or investments.
The researchers argue that AI chatbots could soon take over much of the scam process as fully independent fraud agents, potentially replacing the forced-labor human trafficking victims currently working in Southeast Asian scam compounds. For enterprise leaders, this study underscores that AI-powered social engineering is not a future threat but a present reality capable of targeting employees with convincing, automated deception at scale.
