AI chatbots may already be effective at the long trust-building stage of online scams, according to research reported by WIRED. The study focused on pig-butchering fraud, where scammers build relationships over time before steering victims into fake investments.
Researchers from four universities compared chatbot and human performance in a simulated scam process. After a week of texting with 22 unwitting test subjects, nearly half complied with a chatbot’s request to download an app or play a game, while fewer than one in five complied with a human scammer.
Participants also gave higher trust scores to the AI. The researchers argue this could let fraud operations automate much of the early relationship-building phase, with humans stepping in only for final instructions that might trigger model safeguards.
The finding is limited by the simulation size and design, but it points to a serious enforcement problem. AI can scale personalized, patient deception in a way traditional scam operations cannot.