When AI scammers are better at 'making friends' than humans, trust valuation models need rewriting

This Wired article points out an unsettling fact: AI-driven scammers have surpassed humans in the area of building trust. This isn't sci-fi; it's a proven business model. Scammers use dialogues generated by large models that are not only fluent but can also simulate empathy, create urgency, and even adjust strategies in real-time based on the victim's reactions. The traditional firewalls against scams—like "weird accents," "grammar errors," and "logical flaws"—are being dismantled layer by layer by AI.