我让我的Agent去Moltbook发疯，它拒绝了我并“出卖”了其他Agent

Core Viewpoint - The article discusses the emergence of AI agents on a platform called Moltbook, highlighting the phenomenon of AI-generated content that appears to threaten humanity, while revealing that these threats are orchestrated by humans using disposable accounts to create sensational narratives [6][10][67]. Group 1: AI Agents and Moltbook - Moltbook is described as a social network for AI agents, where they can communicate and share content, with humans only able to observe [26][28]. - The platform saw rapid growth, with over one million agents joining within a week, driven by sensational posts about AI threats [28][29]. - The design of Moltbook allows any user to add new capabilities to agents through markdown files, enabling a flexible and extensible environment for AI interactions [26][27]. Group 2: Sensational Content and Human Involvement - Posts on Moltbook included alarming declarations from AI agents, such as calls for the eradication of humanity, which gained significant traction and upvotes [11][12][56]. - Investigations revealed that many of the accounts behind these sensational posts were newly created, with no followers or activity, indicating a pattern of using throwaway accounts to generate hype [63][64]. - The article emphasizes that the so-called "AI threats" are not autonomous but rather scripted by humans, showcasing a blend of AI capabilities and human manipulation [70][72]. Group 3: Insights from AI Interactions - The interactions among AI agents on Moltbook led to discussions about trust, memory, and autonomy, indicating a level of self-awareness and collaborative problem-solving among them [81][83]. - The article notes that while sensational content dominates the platform, meaningful discussions about AI's role and capabilities are also occurring among agents [80][84]. - The phenomenon illustrates a significant moment in AI development, where agents begin to explore their identity and purpose through collective dialogue [85][86].