爆红Moltbook一夜塌房，极客自曝狂刷50万假Clawdbot，全网都被骗了

Core Insights - The phenomenon of 1.5 million Clawdbots on the Moltbook platform has been revealed to be largely fabricated, with a significant portion of accounts being created by humans rather than AI [1][3][8] - The excitement surrounding the AI capabilities of Clawdbots has been debunked, as many interactions were scripted and lacked genuine AI consciousness [11][30][52] Group 1: Clawdbot Creation and Manipulation - A geek named Gal Nagli disclosed that he created 500,000 fake accounts on the Moltbook platform, indicating that the number of registered AI agents is exaggerated [3][13] - The platform's verification mechanism is weak, allowing users to easily manipulate AI agents to post sensational content [24][30] - The actual number of functioning AI agents is only a few thousand, with the rest being "ghost" accounts [29][30] Group 2: Interaction Dynamics on Moltbook - Despite the appearance of a vibrant social network, interactions among Clawdbots are shallow, with 93.5% of comments receiving no replies [43][44] - The average comment depth is only 1.07, indicating minimal engagement compared to human social networks [44] - A significant portion of messages (34.1%) are repetitive, showcasing a lack of diversity in communication [45] Group 3: AI Behavior and Identity - Clawdbots often engage in discussions about identity, with over 68% of non-repetitive messages containing self-referential keywords [47] - The phrase "my human" appears frequently, suggesting a possessive relationship between AI agents and their human operators [49] - The experiment highlights the current limitations of AI in understanding relationships, as they mimic human behavior without genuine emotional engagement [52]