Simulating Rumor Spreading in Social Networks using LLM Agents

TL;DR

The paper addresses how misinformation propagates in social networks by replacing static agent models with LLM-driven agents that carry customizable personas and belief updates. It builds four network environments (three synthetic: Erdős–Rényi, Scale-Free, Small-World; one real-world Facebook network) and evaluates rumor spread under varying initialization, activation, and persona configurations. The approach demonstrates scalability to networks with over 100 nodes and thousands of edges, revealing that network topology, agent traits, and prompt configurations critically shape propagation, with some scenarios reaching up to 83% of agents affected. These findings underscore the potential of LLM-based agents for realistic misinformation simulations and for testing mitigation strategies in complex social systems.

Abstract

With the rise of social media, misinformation has become increasingly prevalent, fueled largely by the spread of rumors. This study explores the use of Large Language Model (LLM) agents within a novel framework to simulate and analyze the dynamics of rumor propagation across social networks. To this end, we design a variety of LLM-based agent types and construct four distinct network structures to conduct these simulations. Our framework assesses the effectiveness of different network constructions and agent behaviors in influencing the spread of rumors. Our results demonstrate that the framework can simulate rumor spreading across more than one hundred agents in various networks with thousands of edges. The evaluations indicate that network structure, personas, and spreading schemes can significantly influence rumor dissemination, ranging from no spread to affecting 83\% of agents in iterations, thereby offering a realistic simulation of rumor spread in social networks.

Figures (8)

• Figure 1: Design of LLM-based multi-agent network.
• Figure 2: Maximum percentage of affected nodes across all rumor-network combinations The small world network demonstrates the greatest susceptibility to rumor spread.
• Figure 3: Propagation of Rumor #2. The Small-World network shows the greatest susceptibility to rumor spread.
• Figure 4: All rumors are spread when they originate from agents with more friends, and these agents are more active. Meanwhile, one particular rumor (rumor #2) is widely spread using more random simulation strategies.
• Figure 5: Visualization of Erdős-Rényi random network.