People Attribute Stable Personalities to Chatbots

Even when users know text comes from a chatbot, they infer consistent, humanlike personality traits from its writing; humans detect AI partly by length and formality, and matching those features erodes detection. A controlled study (JGU Mainz, published in Scientific Reports) found humans judged AI authors reliably but relied on surface cues, while AIs were worse at recognizing AI writing. — This changes how we should think about trust, disclosure, therapeutic use, misinformation, and regulation because personality impressions (not factual status) will drive human behavior toward AIs.