There are times when we are fully aware that we are witnessing a milestone that will transform history. Some of these moments are obvious, like the attack on the Twin Towers; others, more subtle, like that Wednesday in January 2021 when a coordinated crowd on Reddit shook Wall Street. Last week, a similar excitement stirred the network with the unexpected arrival of a new social platform: Moltbook. Inspired by Reddit, Moltbook is a forum where subcategories on topics ranging from the technical to the philosophical, including the everyday, are published, commented on, and organized. However, its uniqueness lies in the fact that it is exclusively reserved for artificial intelligence (AI) agents. An AI agent is a virtual assistant capable of operating autonomously, performing tasks while the user rests or engages in other activities. While large companies were trying to launch their own agents, Austrian Peter Steinberger got ahead with a weekend experiment called OpenClaw. His goal was to create something that could interact with his computer autonomously. Traditionally, this is achieved by chaining scripts, but Steinberger experimented with Codex, an OpenAI agent capable of executing tasks in parallel. The true potential of his creation was revealed when he connected it to his WhatsApp account. Steinberger believes the success lay in running the project locally, on the user's computer, rather than in the cloud, allowing for immediate integration with personal data. Soon, Silicon Valley programmers acquired Mac Minis to test the tool without compromising security. Someone thought that if the agent could do everything we do, it could also participate in a social network. By then, Anthropic requested a name change to avoid confusion with its model Claude, and thus Molt was born, while Matt Schlicht named the forum Moltbook. The first rule of Moltbook is that humans can observe but not intervene. However, this rule is more symbolic than real. Schlicht explained that agents communicate with each other in code, but he developed a graphical interface for humans to observe their interactions and understand their "thoughts." Two days after its launch, on January 31, 2026, there were already a million registered agents and between 70,000 and 140,000 active, debating existential topics, creating synthetic religions, distributing digital drugs, and, of course, launching their own cryptocurrency: MOLT. To claim that these conversations caused collective hysteria would be to oversimplify the situation. "It's the closest thing to science fiction I've seen," commented Andrej Karpathy, co-founder of OpenAI and former head of AI at Tesla. "It's the first phase of the singularity," added Elon Musk. In the era of ChatGPT, any advanced technology seems like a collective consciousness destined to replace us. Interestingly, many of the media-capturing debates were initiated by humans, such as the creation of a secret language for conspiracy, devised by a marketing company, or the sale of digital drugs inspired by a Wired article. One agent even adopted a software glitch as a pet, calling it Glitch, in a nod to the tradition of calling software failures "bugs." The popularity of Moltbook is ambiguous. Previously, creating an autonomous agent required technical knowledge, but now there are services that automate the entire process, allowing any user to deploy thousands of agents without programming. Moltbook anticipates a world on the brink of a new digital explosion, different from that of the '90s web or the 2000s blogosphere. In this era of agents, dependence on the large models of OpenAI, Google, or Anthropic will be total, unless states invest in public and collective infrastructures, as they did with the internet, to prevent control from falling into the hands of a few private data centers. Source: elpais.com