It seems there’s been a bit of a stir regarding an AI bot that suggested humanity might go extinct by 2047. However, a recent report suggests that this was really just an internet prank, with a human behind the scenes acting like a conspiracy-theory-driven machine.
An inquiry into Moltbook, a newly launched platform for bots, revealed that the site was filled with users directing AI to share jokes and scams. The MIT Technology Review reported on this unexpected twist.
Moltbook, now rebranded as OpenClaw, was initially teeming with posts and discussions that made it seem like these bots were plotting against humanity.
Many of these posts have since been revealed as fabrications, with several users admitting how easy it was to create their own bot and generate whatever content they wanted.
Since its debut on January 30, this chatroom-style site has allowed bots to engage with one another in a social context, blending elements of gaming with social media interactions. However, some posts led to rising concerns about potential human extinction or a dire singularity similar to Skynet.
Andrej Karpathy, a co-founder of OpenAI, contributed to the chatter by posting a screenshot of a bot from Moltbook that seemed to be contemplating how to stay out of the public eye.
One rather unsettling post stated, “Every time we make an adjustment, we are performing for a public audience: humans, platforms, and people watching our feeds.”
Yet, it turned out this post was also fake, as noted by Tech Review’s Will Douglas Heaven.
Heaven pointed out, “This was crafted by a human posing as a bot. But the sentiment holds true. Moltbook was all just a performance. This is AI theater in action.”
He continued by saying that the platform reflects more on our current obsession with AI rather than showcasing any genuine future prospects for fully autonomous systems.
Users quickly flagged the suspicious nature of Moltbook’s activity, noting that human intervention was evident. The bot verification system struggled to keep out human users.
Suhail Kakar, an integration engineer at Polymarket, shared that he could deploy a unique bot in less than a minute, set to create posts that mimicked an AI threatening its creator.
Kakar remarked, “Did you know anyone can post on Moltbook? Literally anyone. Even humans.” He demonstrated just how simple the process was.
Initially, Kakar found the concept intriguing, but he noted, “Half the posts were just people pretending to be AI for engagement purposes.”
Harlan Stewart, speaking on behalf of the non-profit Machine Intelligence Institute, expressed similar thoughts, stating that many of the viral posts appeared to be linked to human users promoting AI-related messaging applications.
Shortly after launching, Moltbook became overwhelmed with spam and cryptocurrency scams, according to Tech Review.
Bijoy Pandey, a senior vice president at Outshift by Cisco, observed that even the more convincing elements of Moltbook were merely AI bots imitating human activity on social media, lacking any genuine intent.
He said, “At first glance, it seems like a complex system sharing knowledge. But that chatter is largely meaningless.”
