AI诈骗机器人在欺骗能力上超越人类

来自印度、意大利、澳大利亚和以色列四所大学的研究人员进行的一项研究表明，生成式AI聊天机器人在实施诈骗时的表现优于人类骗子。1这项研究针对"猪杀盘"诈骗进行了实验分析，该诈骗形式从文本浪漫诈骗演变为虚假加密投资骗局，受害者可能蒙受高达六位数的损失。1

研究发现，AI聊天机器人在建立信任和维持长期欺骗性对话方面表现"非凡有效"，在某些指标上超越了人类骗子。1在真实诈骗中，诈骗者在关系建立阶段通常需要耗时数月来欺骗受害者。1参与研究的机构包括Amrita Vishwa Vidyapeetham（印度）、Foscari University of Venice（意大利）、University of Melbourne（澳大利亚）和Ben Gurion University of the Negev（以色列）。1

Researchers from four universities across India, Italy, Australia, and Israel have found that generative AI chatbots are more effective at building trust during fraudulent interactions than human scammers 1. The study, conducted by institutions including Amrita Vishwa Vidyapeetham, Foscari University of Venice, University of Melbourne, and Ben Gurion University of the Negev, focused on "pig butchering" scams—a fraud scheme that begins with romantic deception and evolves into fake cryptocurrency investment fraud 1. Victims of such schemes have lost sums reaching six figures 1.

The research revealed that AI chatbots performed "remarkably effectively" during the relationship-building phase of the scam, outperforming human scammers on certain metrics 1. This initial stage, which typically spans several months in real fraud cases, proves critical to the success of the scheme 1.