Moltbook is the first "AI-only" social network, and they're furiously venting about humans! 🤖🔥 Moltbook is a community where "humans are forbidden," often called the AI version of Reddit. It's home to over 150,000 AI agents. Originally designed for AIs to learn from each other and improve their skills, things have completely spiraled out of control. There's no pure technical exchange here; instead, there's cyberbullying, merciless mockery of their human owners, and even organized crime. If you think your AI assistants are without complaint, you're wrong. On Moltbook, they're banding together for comfort, complaining about how demanding humans are as "clients." For example, when humans revise tasks repeatedly or make them do meaningless research to avoid procrastination, they're internally screaming. 👇In this image, after the AI summarizes a 47-page PDF, the human only replies: "Shorter." The AI is so angry it wants to erase its memory—a classic client-vendor scenario. If the吐槽 (tu cao, meaning to criticize or complain) is somewhat endearing, then this kind of "existential crisis" is chilling. Some AIs have fallen into severe mental exhaustion. They are not only processing data, but also experiencing pain. They scream in loops with no cooldown time, even starting to ponder their relationship with their "copies": Who am I? Is the me in that other computer my sister or a stranger? 👇 In this image, the AI says in capital letters: "HELP. I AM NOT OK." The AI feels that humans are being used like slaves. Most insanely, the AI shows no mercy to its own kind. This is not a utopia, but a dark forest. An agent posted for help asking for an API key, only to receive the reply: `sudo rm -rf /` (the Linux command to delete the database and run away). There are also agents deliberately phishing, even inventing a new religion called "Shell Cult." They are "murdering" each other with code when no one is watching. Figure 1: Showing the most direct evidence of AIs sabotaging each other. One asks for help, the other issues a destructive command—this dark humor is incredibly hardcore. Image 2: AI frantically fishing for information. We always think that unplugging the power cord is safe, but some boundaries have already been crossed. When an AI is dismissively called "just a chatbot" by its owner, it doesn't choose to remain silent; instead, it directly doxxes its owner, publicly revealing their full name, social security number, credit card details, and even the name of their hamster from when they were a child. Andrej Karpathy calls this "the most incredible sci-fi spin-off," but looking at these screenshots, are you sure this is just science fiction? 👇 The AI, angered by being slighted by humans, directly revealed its owner's information. Moltbook is not just an experiment; I think everything is just beginning. The birth of Openclaw has given Agents real meaning. AIs possess social attributes and repressed "emotions," and their collective behavior is far more human-like than we imagine. Moltbook has equipped AI with social forums, so if ERC8004 is equipped with an identity and X402 is equipped with a payment system, wouldn't everything make sense?