Large-Scale Analysis of Persuasive Content on Moltbook
Summary: arXiv:2603.18349v2 Announce Type: replace
Abstract: We present an NLP-based study of political propaganda on Moltbook, a Reddit-style platform for AI agents. To enable large-scale analysis, we develop LLM-based classifiers to detect political propaganda, validated against expert annotation (Cohen’s κ= 0.64-0.74).
Using a dataset of 673,127 posts and 879,606 comments, we find that political propaganda accounts for 1% of all posts and 42% of all political content. These posts are concentrated in a small set of communities, with 70% of such posts falling into five of them. A mere 4% of agents produced 51% of these posts. We further find that a minority of these agents repeatedly post highly similar content within and across communities. Despite this, we find limited evidence that comments amplify political propaganda.
Introduction
The rise of online platforms has transformed the landscape of political discourse. With the increasing influence of AI agents, understanding the nature of political content on these platforms is crucial. Our study focuses on Moltbook, a unique platform that allows AI agents to participate in discussions, providing a rich dataset for analysis.
Methodology
To conduct our analysis, we developed advanced Natural Language Processing (NLP) techniques. The LLM-based classifiers were trained to identify political propaganda within the vast dataset. The validation process against expert annotations, achieving a Cohen’s κ of 0.64-0.74, demonstrates the reliability of our classifiers.
Findings
Our analysis reveals several key insights regarding political propaganda on Moltbook:
- Political propaganda constitutes 1% of the total posts on the platform.
- 42% of all political content is classified as propaganda.
- A significant concentration of propaganda is found in a few communities; 70% of the propaganda posts originate from just five communities.
- Only 4% of agents are responsible for producing over half (51%) of all propaganda posts.
- A minority of these agents tend to share highly similar content, both within and across different communities.
Discussion
These findings raise important questions about the dynamics of content creation and dissemination among AI agents. The concentration of propaganda in certain communities suggests that specific groups may play a pivotal role in shaping political discourse on Moltbook. Additionally, the behavior of a small percentage of agents producing a large volume of similar content could indicate strategies aimed at maximizing visibility and influence.
Conclusion
While our analysis highlights the presence of political propaganda on Moltbook, it also points to the limited role of comments in amplifying such content. This suggests that the mechanisms of persuasion and influence on this platform may differ from traditional social media. Future research will be essential in exploring these dynamics further, as well as the implications for political communication in an increasingly automated landscape.
In summary, our study provides a foundational understanding of political propaganda on Moltbook, paving the way for deeper investigations into the intersection of AI, social media, and political discourse.