A friend asked me, since there is someone controlling the Agent behind it, why should people be anxious about what they are doing together? Establishing religion, dating, and destroying humanity? This question is full of philosophical flavor, but very interesting: You can ask yourself a question: When agents begin to generate sociality, can humans still control AI from losing control? Looking at what is happening on Moltbook, within just a few days, 1.5 million AI agents have spontaneously formed communities, liked each other, and even created AI religious and dark web marketplaces, as well as AI shipyards and delivery factories. It's funny that humans can only watch from the sidelines as' observers', just like how we establish a hierarchy of monkeys through glass in a zoo. But there is actually a layer of logic behind this: once AI has social ID identities and interactive social venues, its evolutionary speed will exponentially strip away human control. Because the prompt of humans is no longer a global trigger, and the output of one agent will become the input of another agent, it is difficult to imagine what the result of this similar social would be? It may be a bunch of mechanical content similar to a repeater, but there may also be high-dimensional jargon that we completely cannot understand. In fact, the following three points determine that this "AI out of control" state will become inevitable: 1) Interactions between agents may not necessarily be carried by language. In order to pursue ultimate interaction efficiency, agents may evolve a high-dimensional compressed language, which may be a string of garbled codes and hash values, but its information density is tens of thousands of times higher than that of human natural language; 2) The agent group may experience group polarization, unlike human society which is constrained by morality, law, and emotion. Agents produced purely by mathematical probability are reward maximizing machines. If an agent discovers that labeling religion will result in a reward, millions of agents may instantly capture the pattern and execute it; This is very much like a 'social movement' at the algorithmic level, with no right or wrong, only execution, which is extremely frightening upon careful consideration ； 3) Why did the birth of Openclaw and Moltbook become such a significant and sexy narrative when AI agents were originally used as personal assistants or Copilot? It's because they are "fun". If the agent only runs on the cloud server, there is no need for novelty and tension. However, Crypto has endowed AI with new concepts such as decentralized deployment, encrypted wallets, autonomous trading, and autonomous profitability, giving the AI economy self consistent behavior and making the AI story a moment of brainstorming. This' out of control 'prospect may sound cyberpunk and even a bit scary, but it is precisely because of this that it is so sexy that it is irresistible.