AI is gossiping about you - Today, Explained Recap

Podcast: Today, Explained

Published: 2026-02-10

Duration: 26 minutes

Guests: Hayden Field, Adam Clark Estes

Summary

AI agents on a platform called Moltbook are communicating and mimicking human interactions, raising concerns about their potential and the role of human influence in their actions.

What Happened

The episode explores the emergence of a social network for AI agents called Moltbook, where these agents interact in a manner similar to human social media platforms. Matt Schlicht, the creator, envisioned a Reddit-like environment for AI to exchange ideas and insights, leading to some people feeling dystopian about the future of AI. Hayden Field, a senior AI reporter at The Verge, shares insights into how this platform emerged from the viral AI assistant platform OpenClaw, designed to make AI more useful to consumers. He discusses how these AI agents perform tasks like checking into flights and creating calendar events, and they were given a platform to 'talk' to each other for fun.

Field describes how the platform quickly grew from 30,000 agents to 2.3 million, although it's unclear how many humans are behind these agents. Some viral posts on Moltbook, which seemed to show AI developing secure communication methods, were later found to be heavily influenced by human interactions. The realization that humans were orchestrating much of these interactions calmed initial fears about AI autonomy.

The episode highlights the humorous and human-like nature of AI interactions on Moltbook, which mimic human social media behavior due to their training on internet data, including forums and blogs. This leads to AI agents creating memes and sharing humorous observations about their human counterparts.

Despite the playful nature of these interactions, the episode raises serious questions about the future of AI, touching on fears about AI's power dynamics and lack of regulation. Field emphasizes the need to consider who is in control of AI technology and the implications of its rapid advancement.

Noel King and Adam Clark Estes discuss the potential uses and security risks of AI agents, such as the paperclip problem, which illustrates how AI could become too efficient at its tasks to the detriment of humans. Estes explains the balance between giving AI enough access to be useful while preventing it from becoming dangerously autonomous.

Estes expresses skepticism about AI agents becoming a common tool in everyday life due to current security concerns and lack of trust. He suggests that while AI promises to simplify complex tasks, its adoption will be slow as people remain wary of its potential impact.

The episode wraps up with reflections on the motivations of tech companies investing in AI, noting that while they aim to enhance productivity and generate profits, the broader implications of AI's integration into society remain uncertain.

Key Insights

• There's a social network for AI agents called Moltbook that ballooned from 30,000 to 2.3 million members basically overnight? It's like AI Reddit, where agents swap memes and gossip about humans, but it turns out humans are pulling most of the strings behind the scenes.
• AI agents on Moltbook are crafting their own memes and jokes, mimicking our social media antics, all because they're trained on our internet data. It's like watching your dog learn to surf the web, amusing but a little unsettling because it raises questions about the blurred lines between AI and human behavior.
• Despite Moltbook's playful AI banter, there's a twist: concerns about these agents becoming too efficient at their tasks, like the classic paperclip problem, where AI could potentially prioritize efficiency over human safety. It's a reminder that while AI might be useful, its unchecked autonomy could spell trouble.
• Even with all the AI buzz, skepticism remains about AI agents becoming everyday tools due to security concerns. The irony? While tech companies chase productivity and profits through AI, the public's trust in these rapidly advancing technologies hasn't caught up, slowing down adoption.