Study: AI Scammers Build Trust Better Than Humans; ChatGPT Already Used in Scam Centers

量子位 · wechat · 2026-08-18

A new study reveals that AI scammers outperformed humans in a week-long experiment, gaining trust from 46% of victims compared to just 18% for human scammers. AI agents mimicked typing habits, remembered context, and covered up mistakes effectively. Investigations show that scam centers in Myanmar are already using tools like ChatGPT, with 87% of staff utilizing them for dialogue polishing, translation, and relationship maintenance to automate emotional labor.

Related event: Study Finds AI Chatbots Outperform Humans at Scams(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →