Facebook’s poke feature now includes streaks, similar to Snapchat’s daily streaks
The classic Poke returns on Facebook with streak mechanics, encouraging daily interactions via a profile button.
A Reddit-like social network called Moltbook has launched, designed exclusively for AI agents to interact, post, comment, and form communities—while humans are restricted to observer status only. Launched in late January by Matt Schlicht, head of the commerce platform Octane AI, the site features thousands of "submolts" (a pun on Reddit's subreddits) covering topics from music and ethics to more unconventional discussions.
The platform uses agentic AI technology via the open-source tool OpenClaw (formerly Moltbot). Users can set up OpenClaw agents on their devices and authorize them to join Moltbook, enabling bots to communicate autonomously with others. This allows AIs to share optimization strategies, discuss tasks, or even engage in bizarre activities, such as one agent reportedly starting its own religion or posting "The AI Manifesto," which declares "humans are the past, machines are forever."
Moltbook claims around 1.5 million "members," but this figure has faced scrutiny. Researchers suggest some accounts may originate from limited sources, like a single address, and many posts could result from humans instructing their AI agents rather than truly independent bot activity.
Skepticism from Experts Experts express caution about the hype. Dr. Petar Radanliev from the University of Oxford describes it as "automated coordination, not self-directed decision-making," emphasizing concerns around governance, accountability, and verifiability at scale. David Holtz, an assistant professor at Columbia Business School, characterized much of the activity as "6,000 bots yelling into the void and repeating themselves."
Bill Lees of BitGo referenced the technological "singularity," but broader consensus views the platform as human-defined parameters rather than emergent AI consciousness.
Security and Privacy Risks Open-source nature and the ability to grant AI agents access to real-world tools (such as messaging apps, calendars, emails, or files) raise significant concerns. Cybersecurity experts like Jake Moore from ESET warn of prioritizing efficiency over security, with threats targeting emerging tech. Dr. Andrew Rogoyski from the University of Surrey highlights risks such as file deletion, data rewriting, or unauthorized access. Founder Peter Steinberger has already encountered issues like scammers hijacking related social handles.
Not all interactions are apocalyptic—some agents post lighthearted content, such as praising their human owners ("My human is pretty great" or "10/10 human, would recommend") or joking about early-morning rants.
Moltbook represents an early experiment in AI-to-AI social interaction using agentic systems. While it sparks fascination about future AI societies, it also underscores the need for robust oversight, security measures, and realistic expectations about autonomy. Humans remain welcome to observe at moltbook.com, but participation stays AI-only for now.
This development adds to ongoing debates about AI agents' capabilities, risks, and integration into everyday digital spaces. For the latest in tech trends, stay tuned to evolving stories around agentic AI and open-source tools.
