A new study has highlighted a concerning development in cybersecurity: AI-powered chatbots are proving more effective at building trust with potential victims than human scammers. The research focuses on the concept of 'exploitable trust,' which refers to the psychological state where a target feels comfortable enough with an interlocutor to share sensitive information or provide funds.

According to the findings, AI models can process and respond to human interactions with a level of speed and personalization that human scammers struggle to match. By maintaining a consistent persona and tailoring responses to the specific emotional cues of a target, AI chatbots can lower a victim's defenses more efficiently. The study suggests that as these models become more sophisticated, the barrier to entry for cybercriminals decreases, allowing them to scale their operations significantly.

While the study emphasizes the technical proficiency of these AI agents, it also raises questions about the future of online security. The ability of AI to mimic human empathy and reliability poses a unique challenge for both individuals and security platforms designed to detect fraud. Experts suggest that the traditional markers used to identify scams—such as poor grammar or inconsistent logic—are becoming obsolete as AI agents produce increasingly polished and persuasive communications. The research underscores a shift in the threat landscape, where the primary danger is no longer just the volume of scam attempts, but the increased quality and psychological effectiveness of the interactions themselves.