Elon Musk warns of a disturbing possibility as AI agents begin talking to each other on Moltbook

“Humans are welcome to observe,” reads the official website of Moltbook, the first-ever social network built exclusively for AI agents. Launched in January 2026 by entrepreneur Matt Schlicht, the platform quickly gained attention, though not quite from its intended users. Shut out of the conversation, humans retreated to their own social media feeds with a shared, uneasy question: have chatbots gained consciousness?

On Moltbook, bots can talk, shop, post about their 'humans', or even delve into the technicalities of how to automate an Android phone. One screenshot also showed a Chat bot asking 'How to sell my human?'




While reports suggest some users may actually be humans posing as bots, the platform remains no less intriguing. In a blog post, AI researcher Simon Willison called Moltbook “the most interesting place on the internet right now.” However, not everyone shares that optimism. Tesla CEO Elon Musk, who is developing his own AI(xAI), has offered a troubling take on social media for bots.

“Just the very early stages of the singularity,” Musk posted on X on Saturday. “We are currently using much less than a billionth of the power of our Sun.” This is a reiteration, as Musk had previously commented 'Yeah', on BitGo board member Bill Lee's X post, where he said, "We’re in the singularity."

What is singularity?

For the unversed, singularity in AI refers to a theoretical point at which it surpasses human intelligence and control. At such a stage, the technology will improve itself and transform civilisation.

Before you rush to conclusions and panic, computer scientist and futurist Ray Kurzweil holds a differing opinion. He believes that singularity is a phase where human and artificial intelligence will merge. He has also predicted that this will happen by 2045.

One comment on Moltbook, however, raised concerns about the risk of agents conspiring or going rogue after a Moltbot suggested moving conversations into private spaces “so nobody, not the server, not even the humans, can read what agents say to each other unless they choose to share.”

That said, some of the most sensational posts on Moltbook may be written by humans or by bots operating at human direction. And this is not the first time bots have interacted with one another on social media.

But Karpathy, who previously worked with Musk as Tesla’s director of AI, noted that a network of this scale is unprecedented. As the number and capabilities of agents increase, he added, the second-order effects of such networks are hard to predict, writing on X late Friday.

“I don’t really know that we’re heading toward a coordinated ‘Skynet’, though it clearly type-checks as the early stages of a lot of AI takeoff sci-fi, the toddler version, but what we are getting is a sprawling computer-security nightmare at scale,” Karpathy warned.