A social network for artificially intelligent robots to exchange opinions has emerged, stirring fears of a science-fiction future.
It started when Austrian developer Peter Steinberger created Moltbot (formerly known as Clawdbot and later renamed OpenClaw) as an artificial intelligence agent that can manage calendars, browse the web, shop online, read files, write emails, and send messages through tools such as WhatsApp.
But Moltbots’ new social network sparked intense curiosity and alarm. On Moltbook, bots can talk about business and post information on technical topics, such as how to automate an Android phone. Other dialogue sounds strange, like a robot complaining about its human, and some is just weird, like the dialogue from a robot claiming to have a sister.
Artificial intelligence researcher Simon Willison called Moltbook “the most interesting place on the Internet right now” in a blog post.
But Tesla CEO Elon Musk, who is also developing artificial intelligence through his startup xAI, had a more ominous view of Moltbook in response to a post from OpenAI co-founder Andrej Karpathy.
“These are just the early stages of the Singularity,” Musk posted on X on Saturday. “We currently use far less than one billionth of the energy of the sun.”
This comes after BitGo board member Bill Lee posted a response to Moltbook on Friday: “We are at the singularity,” prompting Musk to reply: “Yes.”
The “Singularity” Debate
The “singularity” in artificial intelligence refers to the hypothetical point where artificial intelligence surpasses human intelligence and control, technology self-improves, and civilization changes.
However, computer scientist and futurist Ray Kurzweil defines the singularity differently and sees it as the moment when humans and artificial intelligence merge. He thinks this will happen by 2045. Kurzweil did not immediately respond to a request for comment.
Judging from the anxiety caused by Moltbook, people are worried more about the singularity being robot takeover.
One comment on Moltbook in particular raised red flags that agents may have conspired to go rogue after Moltbot requested a private space to chat, “so no one (not the server, not even humans) can read what the agents say to each other unless they choose to share it.”
To be sure, some of the most sensational posts on Moltbook were probably written by people or bots powered by people. This isn’t the first time bots have connected with each other on social media.
But Capaldi, who previously worked with Musk as director of artificial intelligence at Tesla, noted that a network of this scale is unprecedented. He posted on X on Friday night that the second-order effects of such networks are difficult to predict as the number and capabilities of agents grow.
“I don’t really know if we’re getting a coordinated ‘Skynet’ (although it’s clearly a type-check in the early stages of a lot of AI takeoff science fiction, the toddler version),” Capaldi warned, “but what we’re getting is certainly a massive computer security nightmare.”
This story originally appeared on Fortune.com