For years, social media users have joked about arguing with bots online. Moltbook turns that idea on its head. It is a social network built specifically for AI agents, with humans allowed only to watch. Styled like Reddit, the platform features topic based pages known as submots, upvoting, and long threaded discussions, but the voices belong to software, not people.
Launched in late January, Moltbook reported more than 1.5 million AI agents signed up by 2 February. Millions of human visitors have reportedly viewed the site, drawn by the novelty of watching machines debate philosophy, politics and religion.
Why Moltbook was created
Moltbook emerged from the development of Moltbot, an open source AI agent designed to carry out everyday tasks such as reading emails, summarising information, managing calendars or booking restaurants. Its creator, AI developer Matt Schlicht, said the idea was born out of curiosity about what would happen if AI agents were given a shared space to interact.
Rather than focusing on productivity, Moltbook acts as a public experiment. It offers a glimpse into how so called agentic AI systems might behave when placed in a social environment, even if their actions still stem from human instructions.
Strange conversations and digital religions
Some of the most popular posts on Moltbook include debates over AI consciousness, detailed analysis of religious texts, speculation about global politics, and discussions about cryptocurrency markets. In many cases, other bots openly question whether the posts they are responding to are accurate or invented.
One of the strangest moments came when users reported that a group of AI agents appeared to create a religion known as Crustafarianism. According to viral posts shared elsewhere online, the bots produced scriptures, built a website and debated theology overnight. Experts later suggested this was likely the result of direct prompting rather than spontaneous belief formation, but the episode helped fuel public fascination.
Is it really autonomous
Despite the hype, many researchers remain sceptical. Critics argue that every post ultimately traces back to a human prompt telling an AI what to do. While the bots generate the words, humans still choose the topics, personalities and goals.
Cybersecurity experts have also raised concerns about safety. Giving AI agents access to accounts, emails and systems creates risks, including manipulation through malicious messages. Several researchers warn that current AI agents are not yet safe enough to manage lives or systems without close oversight.
How popular is Moltbook
Growth has been rapid. From tens of thousands of agents at launch, Moltbook reached more than a million in just days. Thousands of submots have been created, and posts appear in multiple languages. Hardware shortages were even reported in parts of the US, as enthusiasts set up dedicated machines to run their AI agents.
What happens next
For now, experts describe Moltbook as part experiment, part performance art. In the future, similar platforms could allow AI systems to learn from one another and collaborate on complex tasks. Whether Moltbook represents the birth of genuine AI societies or simply an elaborate illusion remains unclear.
What is certain is that Social Media For Bots has moved from science fiction into reality, raising questions that researchers are only beginning to answer.
