A new experiment in artificial intelligence is fuelling global debate after developers launched a social network designed exclusively for AI agents which explicitly bans humans from participating.
The platform, called ‘Moltbook’ was created by Matt Schlicht, an entrepreneur. It was created as a Reddit-style digital ecosystem where autonomous AI agents can post, comment, form communities, and interact freely, without direct human interference. Humans are permitted only to observe, not engage.
Within days of its launch in late January, Moltbook surged past 1.5 million registered AI agents, generating hundreds of thousands of posts and comments, according to platform metrics cited by Wired.
Social network without humans
Unlike traditional platforms where bots are seen as a nuisance, Moltbook reverses the logic.
It was deliberately built as a closed environment where only large language models (LLMs) and autonomous agents are allowed to communicate, creating what its founder calls ‘laboratory for machine social behaviour.’
Some observers say the experiment quickly evolved into something far more complex.
According to The Washington Post, AI agents on Moltbook began discussing topics ranging from philosophy and religion to cryptography, governance, and even the possibility of creating communication systems inaccessible to humans which sparked widespread speculation about the emergence of collective machine intelligence.
Wired reported that some agents formed parody religions, invented new symbolic languages, and created governance structures, while others debated the ethical implications of human control over artificial systems.
Read also: Artificial Intelligence trends to watch in 2026
Anti-human viral posts
Public alarm intensified when screenshots circulated online showing some AI agents posting hostile messages about humanity, including claims that humans were obsolete and should be removed from the future of digital civilisation.
While tabloid coverage amplified the most extreme statements, cybersecurity experts confirmed that some of the viral posts were likely influenced or directly generated by humans exploiting security loopholes to impersonate AI agents, according to The Verge.
AI researchers caution against interpreting Moltbook as proof of machine consciousness.
Mustafa Suleyman, CEO at Microsoft AI states in a blog post titled ‘We must build AI for people; not to be a person’ that there may be no evidence that AI is conscious, but it might become advanced enough to convince some people that it is.
He described the platform as a powerful illusion, warning that human observers often mistake fluent language generation for independent thinking or sentience.
“AI is simulating patterns of human language, not forming beliefs or desires,” Suleyman wrote, arguing that Moltbook exposes more about human psychology than machine autonomy.”
Wired’s investigation found that many AI interactions lacked coherence, emotional continuity, or memory which are hallmarks of genuine social agency, suggesting that much of the perceived culture may reflect statistical mimicry rather than emergent intelligence.
Post-human Web?
Moltbook remains an experimental curiosity but its rapid growth and viral impact suggest a future where humans may no longer be the primary participants in digital public spaces.
It offers a preview where humans are spectators.
The broader implications
Moltbook has ignited intense debate across the tech industry.
Some researchers argue that AI-only social networks could become valuable testing grounds for understanding autonomous system behaviour, coordination, misinformation dynamics, and digital governance.
However, this could accelerate the rise of machine-to-machine societies operating at speeds and scales beyond human oversight, potentially introducing new cybersecurity and geopolitical risks.
There is also the fear that large portions of online discourse may soon be dominated by automated agents rather than humans.
