Social Behavior Among Autonomous AI: How Large Language Models Interact in Dynamic Networks

Narges Fardnia; Fatemeh Seyedin; Matthias Becker; Mahmoudreza Babaei; Adrian Weller

cs.SIAug 25, 2026

Social Behavior Among Autonomous AI: How Large Language Models Interact in Dynamic Networks

Authors: Narges Fardnia, Fatemeh Seyedin, Matthias Becker, Mahmoudreza Babaei, Adrian Weller

Organizations: Gisma University of Applied Sciences, BRAIN Research Center, Potsdam, Germany · Leibniz University Hannover, Germany · University of Cambridge, Cambridge, United Kingdom

Abstract

Cooperation is a cornerstone of human societies, enabling collective progress in dynamic and uncertain environments. With the advent of AI systems acting autonomously, it becomes crucial to understand not only human-AI cooperation but also AI-AI interactions in adaptive networks. In this work, we examine the interactions of AI using Large Language Models -- Mistral, Llama3, Gemma3, and Phi3 -- in a public goods game within dynamic network structures. Our experiments were conducted under single-model and mixed-model conditions across Watts-Strogatz (WS), Barabasi-Albert (BA), and Erdos-Renyi (ER) networks. We analyzed the impact of model architecture, network topology, and prompt design on cooperative behavior. Results show that Mistral and Llama3 offer high cooperation rates, while Phi3 shows defective tendencies. Additionally, the random structure of Erdos-Renyi networks dramatically improves cooperation. Prompt design also plays a key role; a society-benefits prompt leads to a higher cooperation level. These findings offer a preliminary framework for LLM-based simulations in adaptive social networks.

Figures & tables

Panel of Figure 1 — Figure 1: Action Stage: Deciding Whether to Cooperate or Defect. (a) This Snapshot Shows a Part of the Network Where One AI Sits at the Center, Connected to Four Neighboring AIs. (b) The Central AI Chooses to Cooperate, Contributing $x\$$ to Each of Its Four Neighbors. As a Result, Its Own Budget Drops to$ (B_{1}-4x)$ $. (c) At the Same Time, Three Neighbors Decide to Cooperate, While One Chooses to Defect. The Defector Contributes Nothing but Still Receives$ 2x$ $from Each Cooperator—Earning a Total of$ 6x$ $and Ending Up with$ (B_{1}-4x+6x)$$ .

Panel of Figure 2 — Figure 2: Rewiring Stage: Deciding Whether to Add or Remove the Connections.

Table 3

Panel of Figure 3 — Figure 3: Mixed-Model Setup: (a) Degree Evolution over Rounds Across Different Network Structures. (b) Clustering Coefficient Evolution over Rounds Across Different Network Structures.

Figure 4: Degree Evolution over Rounds in the Single-Model Experiment

Figure 5: Clustering Coefficient Evolution in the Single-Model Experiment

Multi-LLM Collaboration

Multi-LLM collaboration studies how large language models share work, exchange information, and combine outputs to improve capability, reliability, or cost.

See all 85 papers

The Butterfly Defects: Tracing Cooperation in LLM Agent Networks
Oct 23, 2026Maria Sampani, Maria Lymperaiou, Angeliki Dimitriou +3Multi-LLM Collaboration Multi-Agent Coordination
National Technical University of Athens
Collective cooperation without individual fidelity in LLM agents
Jun 29, 2026Henrique Ferraz de Arruda, Carlos Gracia Lázaro, Alberto Aleta +1Multi-Agent LLM Systems Human Behavior Simulation
ARAID Foundation, Zaragoza, Spain. · Institute for Biocomputation and Physics of Complex Systems, University of Zaragoza, Zaragoza, Spain. · Universidad San Jorge (USJ), Zaragoza, Spain. +1
Learning social norms enhances compatibility in dynamic human-AI coordination
Jul 8, 2026Yi Yang, Siyuan Liu, Xin Gao +4Multi-Agent Coordination Human-AI Interaction
School of Vehicle and Mobility, Tsinghua University, Beijing, China · State Key Laboratory of Intelligent Green Vehicle and Mobility, Tsinghua University, Beijing 100084, China · State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing, China +2

Oct 23, 2026J/K move · Enter open · S save

The Butterfly Defects: Tracing Cooperation in LLM Agent Networks

Maria Sampani, Maria Lymperaiou, Angeliki Dimitriou, Giorgos Filandrianos+2

National Technical University of Athens

Multi-LLM Collaboration Multi-Agent Coordination Prompt Sensitivity

Open paper AskPDF arXiv

As large language models (LLMs) increasingly operate as interconnected agents, a fundamental question emerges: does individual cooperation translate into collective cooperation? We investigate this question through the Iterated Prisoner's Dilemma (IPD), moving from dyadic interactions to homogeneous and heterogeneous networks of open-weight LLMs. Our findings reveal that cooperation is highly sensitive to model composition and seemingly minor linguistic changes. Models that cooperate with identical peers can shift toward defection in mixed populations. A simple uncertainty cue can increase or decrease cooperation, redistribute costs, and affect unexposed agents several hops away. Even reversing action-label order can dramatically alter behavior, with uncertainty framing sometimes amplifying this sensitivity. Together, these findings challenge the notion of cooperation as a stable model-level property, revealing how strategic behavior depends on interaction partners, network positions, and the language used to frame decisions.

Abstract

Multi-LLM Collaboration

Explore similar work

The Butterfly Defects: Tracing Cooperation in LLM Agent Networks

Collective cooperation without individual fidelity in LLM agents

Learning social norms enhances compatibility in dynamic human-AI coordination