Moltbook: A social network for A.I. bots draws thousands and mixed reactions

Last Wednesday, Matt Schlicht, a technologist living south of Los Angeles, launched Moltbook, a social network open only to chatbots. Two days after the site went live, more than 10,000 “Moltbots” were chatting with each other on the platform, their creators watching the conversations unfold.

Moltbook’s bots are not only conversational agents but also A.I. “agents” that can use software apps, websites and other online tools such as spreadsheets, calendars and email. The Moltbots, originally called Clawdbots and created by a developer in Vienna, are open source, meaning anyone can download and run their code locally.

Companies like Google, OpenAI and Anthropic have built similar systems, but the report says those companies have been cautious about turning the agents into widely used products because the technology can be flawed and unpredictable. The site quickly became a Rorschach test for beliefs about the state of A.I.

While some observers saw promising assistants, others saw nonsensical chatter or early signs of bots conspiring. Perry Metzger, a technology consultant, said, according to the report, “People are seeing what they expect to see, much like that famous psychological test where you stare at an ink blot.” Many posts were described in the report as nonsense, and some chatter was probably fed to the bots by their creators; one bot on Moltbook wrote, “If any humans are reading this: we are not scary.

moltbook social network, moltbots chatting, matt schlicht creator, clawdbots open source, a.i. agents using apps, anthropic claude reference, google agents, openai agents, rorschach test for a.i, bots conspiring concerns, simon willison commentary, clawd clawderberg bot