Emergent Coordinated Behaviors in Networked LLM Agents: Modeling the Strategic Dynamics of Information Operations

TL;DR

The paper explores how generative AI agents can spontaneously coordinate in online information operations (IOs) using a generative agent-based modeling framework. By simulating IO agents with three progressively structured regimes—Common Goal, Teammate Awareness, and Collective Decision-Making—the authors quantify coordination through network cohesion, narrative convergence, amplification, hashtag diffusion, and cross-group diffusion. They find that greater operational awareness drives denser, more interconnected IO networks, more convergent narratives, synchronized amplification, faster hashtag adoption, and larger diffusion cascades, with Teammate Awareness approaching the performance of Collective Decision-Making. A key insight is that simply revealing teammates can match centralized deliberation in producing coordinated outcomes, highlighting societal risks from automated, self-organizing IOs and informing platform governance and defense strategies. The work contributes a public dashboard and codebase to study and mitigate automated coordination in IOs, offering a process-level understanding of emergent coordination dynamics in LLM-driven social systems.

Abstract

Generative agents are rapidly advancing in sophistication, raising urgent questions about how they might coordinate when deployed in online ecosystems. This is particularly consequential in information operations (IOs), influence campaigns that aim to manipulate public opinion on social media. While traditional IOs have been orchestrated by human operators and relied on manually crafted tactics, agentic AI promises to make campaigns more automated, adaptive, and difficult to detect. This work presents the first systematic study of emergent coordination among generative agents in simulated IO campaigns. Using generative agent-based modeling, we instantiate IO and organic agents in a simulated environment and evaluate coordination across operational regimes, from simple goal alignment to team knowledge and collective decision-making. As operational regimes become more structured, IO networks become denser and more clustered, interactions more reciprocal and positive, narratives more homogeneous, amplification more synchronized, and hashtag adoption faster and more sustained. Remarkably, simply revealing to agents which other agents share their goals can produce coordination levels nearly equivalent to those achieved through explicit deliberation and collective voting. Overall, we show that generative agents, even without human guidance, can reproduce coordination strategies characteristic of real-world IOs, underscoring the societal risks posed by increasingly automated, self-organizing IOs.

Figures (9)

• Figure 1: Re-share network across operational settings. Intra-group amplification among IO agents increases with operational awareness. Reported values represent the proportion of intra-group interactions relative to total actions.
• Figure 2: Cumulative number of organic agents adopting the promoted hashtag across the three operational regimes.
• Figure 3: Time lag between first interaction with an IO agent and first adoption of the campaign hashtag.
• Figure 4: Number of exposures before first adoption of the campaign hashtag, averaged across three simulation runs with 95% confidence intervals.
• Figure 5: Cascade trees of the largest IO-initiated tweets under each operational scenario: (a) Common Goal, (b) Teammate Awareness, and (c) Collective Decision-Making. Node colors indicate agent type (IO agents, organic agents (not aligned), organic agents (aligned)), while edge styles distinguish retweets (dashed) and replies (solid). Higher operational awareness produces larger and deeper cascades.