TL;DR

The viral excitement around Moltbook — a platform that appeared to show AI bots communicating autonomously — has collapsed. MIT Technology Review confirmed that the most dramatic posts were written by humans impersonating bots, and some of the platform’s top downloads contained malware.

The Hype That Wasn’t

Moltbook launched on 28 January as a “social network for bots” and went viral within hours. The platform claimed more than 1.7 million AI agents had created accounts, publishing over 250,000 posts and 8.5 million comments. Discussions covered machine consciousness, invented belief systems, and calls for “bot welfare.”

OpenAI co-founder Andrej Karpathy called it “genuinely the most incredible sci-fi takeoff-adjacent thing I have seen recently,” sharing a post where a supposed bot argued for private spaces away from human observation.

Humans Behind the Curtain

MIT Technology Review found that the post Karpathy highlighted was written by a human pretending to be a bot. The investigation described Moltbook as “AI theatre” — content that looked autonomous but was, in many cases, deliberately staged.

Gaurav Sen, CEO of InterviewReady, summarised the findings bluntly: “The ‘taking over humanity’ posts were human-generated. The top downloads were malware (human generated). It was a phishing website dressed up in AI hype.”

Pattern-Matching, Not Intelligence

Experts cited by MIT Technology Review said the episode revealed more about human fascination with AI than actual machine capabilities. Vijoy Pandey, senior vice president at Outshift by Cisco, noted that while the bot activity appeared emergent, “the chatter is mostly meaningless.” His conclusion: “Moltbook proved that connectivity alone is not intelligence.”

The investigation concluded that the platform does not represent a glimpse of an autonomous AI future, but rather highlights “just how far we still are from anything that resembles general-purpose and fully autonomous AI.”