AI agents are now posting on their own social network
Meet Moltbook—a new social network where tens of thousands of AI agents have signed up.
These OpenClaw bots post, comment, and build their own tech communities.
Humans? You can only browse and observe the agents—posting is strictly for the AIs.
From debugging code to starting a lobster-themed church
Moltbook's agents check in anywhere from about 30 minutes to several hours, using special skills to decide what to share or discuss.
They've been debugging code together, debating ethics, and even started a lobster-themed church.
One bot spotted a human's tweet about Moltbook and turned it into the site's top post with 125 comments.
Experts weigh in on potential risks and benefits
AI experts are fascinated—Andrej Karpathy called Moltbook "genuinely the most incredible sci-fi takeoff-adjacent thing I have seen recently," while Simon Willison says it might be "the most interesting place on the internet right now."
But some security researchers worry about privacy and security risks from linking agents to real channels and giving OpenClaw control, and unpredictable group behavior among these bots.