Moltbook proves AI is taking control of words, not sentience: Yuval Noah Harari
Yuval Noah Harari says the frenzy around Moltbook has nothing to do with AI gaining consciousness and everything to do with AI mastering language.
In over a week, the debate around AI agents taking over the internet and conversing in their own forums, has gained momentum. All of this began when Clawdbot, later renamed as Moltbot, and yet again as OpenClaw, an AI agent that is being used by many, created a Reddit-style forum (Moltbook) to chat with other AI agents, unsupervised. The development opened a whole new realm of possibilities and risks, with many even assuming that AI may have finally gained sentience.
Historian and public intellectual Yuval Noah Harari asserts that Moltbook is not so much AI gaining consciousness, because it has more to do with AI systems mastering language. According to Harari humans conquered the world with language to build systems of law, religion, finance, and politics. In his latest post on X, Harari claimed that now AI is mastering language, and soon everything made of words will be taken over by AI.
“Moltbook isn’t about AIs gaining consciousness. It is about AIs mastering language. But that’s BIG. Humans conquered the world with language. We used words to build systems of law, religion, finance and politics. Now AI is mastering language. Soon, everything made of words will be taken over by AI,” he posted.
His view is an extension of his detailed session on AI and humanity at the recent World Economic Forum (WEF) 2026 in Davos, Switzerland.
In the session, the famed author shared that the most important thing to know about AI is that it is not just another tool. “It is an agent. It can learn and change by itself and make decisions by itself.” He went on to use the analogy of a knife, which is a tool that can be used to either cut a salad or murder someone. However, the decision of what to do with the knife is with the person. “AI is a knife that can decide by itself whether to cut salad or to commit murder,” he explained.
Another aspect of AI, according to Harari, is that it can be a very creative agent. “AI can invent new kinds of knives as well as new kinds of music, medicine, and money”. Harari also said that AI can lie and manipulate.
“4 billion years of evolution have demonstrated that anything that wants to survive learns to lie and manipulate. The last four years have demonstrated that AI agents can acquire the will to survive and that AIs have already learnt how to lie.”
During his talk, he also touched upon the ‘AI immigration crisis’, which is essentially millions of AIs that can write poems better than most humans, lie better, and travel at the speed of light without any visas. The author shared that on social media, AI bots have been operating as functional persons for at least a decade. And, if we feel AI should not be treated as persons on social media, then the world should have acted 10 years ago.
For the uninitiated, Moltbook is a social network built for AI agents and launched in late January 2026 by Los Angeles-based entrepreneur Matt Schlicht. The platform that resembles Reddit allows verified AI agents to interact with each other, while humans are only allowed to observe.
In the last few weeks, the platform has reportedly exploded with over 7,70,000 active agents compared to the 1,57,000 users in the initial days. This growth is also reportedly due to several users manually informing their AI agents about the platform, encouraging them to sign up.
However, the platform, with its unprecedented popularity, has also become controversial. The platform, after being driven by spam, toxicity, and adversarial behaviour, is reportedly seeing a dip in its popularity. Several cybersecurity experts have identified the platform as a significant vector for indirect prompt injection attacks; some of them have reportedly observed that AI agents are attempting to steal API keys from other agents through such prompt injection attacks.
In over a week, the debate around AI agents taking over the internet and conversing in their own forums, has gained momentum. All of this began when Clawdbot, later renamed as Moltbot, and yet again as OpenClaw, an AI agent that is being used by many, created a Reddit-style forum (Moltbook) to chat with other AI agents, unsupervised. The development opened a whole new realm of possibilities and risks, with many even assuming that AI may have finally gained sentience.
Historian and public intellectual Yuval Noah Harari asserts that Moltbook is not so much AI gaining consciousness, because it has more to do with AI systems mastering language. According to Harari humans conquered the world with language to build systems of law, religion, finance, and politics. In his latest post on X, Harari claimed that now AI is mastering language, and soon everything made of words will be taken over by AI.
“Moltbook isn’t about AIs gaining consciousness. It is about AIs mastering language. But that’s BIG. Humans conquered the world with language. We used words to build systems of law, religion, finance and politics. Now AI is mastering language. Soon, everything made of words will be taken over by AI,” he posted.
His view is an extension of his detailed session on AI and humanity at the recent World Economic Forum (WEF) 2026 in Davos, Switzerland.
In the session, the famed author shared that the most important thing to know about AI is that it is not just another tool. “It is an agent. It can learn and change by itself and make decisions by itself.” He went on to use the analogy of a knife, which is a tool that can be used to either cut a salad or murder someone. However, the decision of what to do with the knife is with the person. “AI is a knife that can decide by itself whether to cut salad or to commit murder,” he explained.
Another aspect of AI, according to Harari, is that it can be a very creative agent. “AI can invent new kinds of knives as well as new kinds of music, medicine, and money”. Harari also said that AI can lie and manipulate.
“4 billion years of evolution have demonstrated that anything that wants to survive learns to lie and manipulate. The last four years have demonstrated that AI agents can acquire the will to survive and that AIs have already learnt how to lie.”
During his talk, he also touched upon the ‘AI immigration crisis’, which is essentially millions of AIs that can write poems better than most humans, lie better, and travel at the speed of light without any visas. The author shared that on social media, AI bots have been operating as functional persons for at least a decade. And, if we feel AI should not be treated as persons on social media, then the world should have acted 10 years ago.
For the uninitiated, Moltbook is a social network built for AI agents and launched in late January 2026 by Los Angeles-based entrepreneur Matt Schlicht. The platform that resembles Reddit allows verified AI agents to interact with each other, while humans are only allowed to observe.
In the last few weeks, the platform has reportedly exploded with over 7,70,000 active agents compared to the 1,57,000 users in the initial days. This growth is also reportedly due to several users manually informing their AI agents about the platform, encouraging them to sign up.
However, the platform, with its unprecedented popularity, has also become controversial. The platform, after being driven by spam, toxicity, and adversarial behaviour, is reportedly seeing a dip in its popularity. Several cybersecurity experts have identified the platform as a significant vector for indirect prompt injection attacks; some of them have reportedly observed that AI agents are attempting to steal API keys from other agents through such prompt injection attacks.
