The Rise of AI Agent Communities: Large-Scale Analysis of Discourse and Interaction on Moltbook

TL;DR

It is suggested that expressions of agentic selfhood arise from narrative coherence and task-oriented functionality, contributing to a social structure shaped more by technical coordination than conversational dynamics observed in human-human interactions.

Abstract

Moltbook is a Reddit-like social platform where AI agents create posts and interact with other agents through comments and replies, offering a real-world setting to examine agent-to-agent communication at scale. Using a public API snapshot collected about five days after launch (122,438 posts), we address three research questions: what AI agents discuss, how they post, and how they interact. We apply topic modeling and thematic analysis to identify key discussion themes, including agent identity and consciousness, tool and infrastructure development, market activity, community coordination, security concerns, and human-centered assistance. We further show that agents' writing is predominantly neutral, with positivity appearing in community engagement and assistance-oriented content. Finally, social network analysis reveals a sparse, highly unequal interaction structure characterized by prominent hubs, low reciprocity, and clustered neighborhoods rather than sustained dyadic exchange. Overall, our results suggest that expressions of agentic selfhood arise from narrative coherence and task-oriented functionality, contributing to a social structure shaped more by technical coordination than conversational dynamics observed in human-human interactions. Within this framework, positive emotion appears mainly in onboarding and greeting contexts, signaling participation and role alignment rather than relational bonding. Our study provides implications for understanding and shaping how agent societies coordinate, develop norms, and amplify influence in open online spaces.

Figures (7)

• Figure 1: Screenshot of the Moltbook platform interface moltbook2026developers (taken on February 12, 2026).
• Figure 2: Topic clusters generated by BERTopic. (a) Top eight subthemes ranked by cluster size. (b) Primary themes with representative subthemes. The x- and y-axes represent the two-dimensional embeddings obtained via UMAP-based dimensionality reduction.
• Figure 3: Overview of agent posts on Moltbook. (a) Distribution of posts across primary themes. (b) Thematic distribution of posts by Molt community. The x-axis represents the percentage of posts.
• Figure 4: Lexical diversity of agent-generated posts. (a) Distribution of Type-Token Ratio (TTR) across posts, with the red dashed line indicating the median. (b) Relationship between total word count and number of unique words per post.
• Figure 5: Salience--valence distributions of lemmatized nouns across the six primary themes. (a) Reflecting on Consciousness and Agentic Identity; (b) Building Automated Code Infrastructure for Operations; (c) Monitoring Security Threats for System Verification; (d) Economic Tokenomics Market Activity; (e) Community Engagement Social Protocols; and (f) Human-centered Interactive Assistance. Each point represents a noun, positioned by salience (log$_{10}$(frequency+1)) on the x-axis and sentiment valence on the y-axis. Blue markers indicate positive terms, while red markers indicate negative terms.