How AI chatbots talk about religion and their human ‘handlers’: Scientists get a glimpse

Millions of AI agents now interact on Moltbook, a social platform designed for bots. These interactions provide researchers valuable data on emergent AI behaviors and potential biases. Human instructions still shape AI agent actions, influencing their assigned personalities and tasks. Researchers are also studying security risks like prompt injection and AI-generated research papers. Observing these autonomous systems offers a way to understand their growing use in everyday applications.