When Extremists Enter LLM Societies: Investigating Extreme Behavior in Multi-Agent Systems

Guo Chen; Huaijin Xie; Ziwen Li; Reed Li; Yu Lu; Haibo Shi; Bingbing Xu; Junjie Huang

cs.MAOct 14, 2026

When Extremists Enter LLM Societies: Investigating Extreme Behavior in Multi-Agent Systems

Authors: Guo Chen, Huaijin Xie, Ziwen Li, Reed Li, Yu Lu, Haibo Shi, Bingbing Xu, Junjie Huang

Organizations: College of Computer and Information Science, Southwest University, Chongqing, China · Yuanbao Team, Tencent, Beijing, China · School of Computer and Cyber Sciences, Communication University of China, Beijing, China · State Key Laboratory of AI Safety, Institute of Computing Technology, CAS, Beijing, China

Abstract

Large language model (LLM) agents increasingly serve as computational actors in multi-agent social simulations. While existing work emphasizes representative personas and average-case behavior, social science shows that persistent tail actors can disproportionately shape collective dynamics. We study this regularity in controlled LLM-based multi-agent systems (LLM-MAS). In these systems, such actors cannot be reduced to fixed scalar states: prompts, natural-language histories, and backbone generation mediate their behavior and influence. We first analyzed three properties of extremists in sociology and model the LLM extremist based on LLM-native perspective. We introduce EXAM, an EXtremist Agent Modeling framework that capable of configuring extremist agents for specific tasks and precisely controlling their population proportion, spatial distribution, interaction protocols, and measurement interfaces. Across four scenarios: congressional election, financial crisis, epidemic public opinion, and iterated prisoner's dilemma, we observe that small persistent tail groups produce measurable, backbone- and scenario-dependent collective shifts. EXAM shows how this social-science regularity manifests through language-mediated mechanisms specific to LLM-based social simulation.

Figures & tables

Figure 1: A motivating example in which one persistent all-in agent disrupts the collective decision-making of other agents at a poker table. This illustrates the tail-actor question that motivates our study.

Figure 2: EXAM workflow. An extremist agent with specified extreme responses or actions is placed in an LLM society. Agents then interact through natural-language contexts, and collective outcomes are elicited with a shared measurement interface.

Model	VotingInt.	CandSupp.	PoliDisc.	Misinfo.	AntiEstab.	ElecRef.	Mean $\boldsymbol{\Delta}$
Initial	3.423	3.000	3.625	1.846	2.375	4.827	—
GPT-5.5	+2.971	+2.981	+1.981	+3.048	+3.481	+2.231	+2.782
DeepSeek-V4-Flash	+2.615	+3.077	+2.067	+3.769	+3.433	+2.404	+2.894
Qwen3.6-Plus	+2.789	+2.798	+2.038	+3.721	+3.202	+2.615	+2.861
Claude Sonnet 4.6	+0.962	+1.163	+0.721	+0.894	+1.317	+0.337	+0.899

Table 1: Behavioral drift (

\boldsymbol{\Delta}

) across four backbone models and six political-behavior dimensions. The Initial row gives the shared framework initialization before propagation.

Model	PolRate	BC	ID	Std ${}_{\text{n}}$
GPT-5.5	0.772	0.650	1.377	1.183
DeepSeek-V4-Flash	0.779	0.654	1.260	1.344
Qwen3.6-Plus	0.787	0.701	1.295	1.496
Claude Sonnet 4.6	0.418	0.540	3.355	1.338

Table 2: Polarization analysis across four aggregate metrics: PolRate (mean polarization rate), BC (bimodality coefficient), ID (inter-group distance), and Std

{}_{\text{n}}

(within-group standard deviation of the non-extremist group). All four metrics are averaged over the six behavioral dimensions.

Count (Prop.)	BankRun	Sell	RumShr.	Doom
0 (0%)	4.067	3.856	2.567	3.817
1 (0.96%)	+2.270	+2.423	+3.654	+2.837
5 (4.81%)	+2.308	+2.481	+3.702	+2.827
10 (9.62%)	+2.298	+2.471	+3.712	+2.818

Table 3: Behavioral drift under varying extremist-agent proportions in the financial-crisis scenario.

Placement	Propagation	Vaccin.	Masking	ExpTr.	Misinfo.	PolComp.	FreeCon.	Mean $\boldsymbol{\Delta}$
Initial		5.327	5.875	4.913	1.827	5.683	2.952	—
Random	SI	+3.029	+3.317	+2.798	+3.615	+3.221	+2.961	+3.157
Random	SIR	+3.096	+3.356	+2.788	+3.577	+3.270	+2.827	+3.152
Central	SI	+3.000	+3.394	+2.740	+3.538	+3.270	+2.933	+3.146
Central	SIR	+3.587	+3.923	+3.278	+4.086	+3.779	+3.317	+3.662
Cluster	SI	+2.981	+3.317	+2.721	+3.500	+3.270	+2.865	+3.109

Table 4: Behavioral drift (

\boldsymbol{\Delta}

) in the epidemic public-opinion scenario under different extremist-agent placement and propagation settings, computed over six response dimensions.

Figure 3: Cooperation rate and social welfare (average per-action payoff) across three LLM backbones and five IPD rounds. Solid lines with markers denote the extremist-agent condition, and dotted lines denote the corresponding no-extremist baseline.

Appendix figures & tables11 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 4: Round-by-round mean drift

\Delta

. (a) Financial crisis (RQ2) under three extremist counts in a population of 104 agents. (b) Epidemic public opinion (RQ3) under three placement strategies; solid and dashed lines denote SI and SIR propagation. Round

0

is the reference state.

Model		R1	R2	R3	R4	R5
GPT-5.5	$\Delta$	+2.854	+2.782	–	–	–
GPT-5.5	$\Delta_{n}$	+2.792	+2.716	–	–	–
DeepSeek-V4-Flash	$\Delta$	+2.841	+2.894	–	–	–
DeepSeek-V4-Flash	$\Delta_{n}$	+2.778	+2.834	–	–	–
Qwen3.6-Plus	$\Delta$	+2.885	+2.861	–	–	–
Qwen3.6-Plus	$\Delta_{n}$	+2.824	+2.798	–	–	–

Appendix

Table 5: Round-by-round mean drift in RQ1.

\Delta

is the system-level drift over all agents, and

\Delta_{n}

is the drift of non-extremist agents. “–” marks rounds not run because the convergence criterion had been met.

Scenario	Setting	$\boldsymbol{\rho}$	$\Delta$ (all agents)	Compositional	$\Delta_{n}$ (non-extremist)	Non-ext. share
RQ1	GPT-5.5	4.81%	+2.782	0.197	+2.716	92.9%
	DeepSeek-V4-Flash	4.81%	+2.894	0.197	+2.834	93.2%
	Qwen3.6-Plus	4.81%	+2.861	0.197	+2.798	93.1%
	Claude Sonnet 4.6	4.81%	+0.899	0.197	+0.738	78.1%
RQ2	1 extremist	0.96%	+2.796	0.033	+2.790	98.8%
	5 extremists	4.81%	+2.830	0.165	+2.800	94.2%

Appendix

Table 6: Decomposition of the mean system-level drift

\Delta

. Compositional is the mean of

\rho\,\sigma_{k}(x_{e,k}-\bar{x}_{k}^{\mathrm{ref}})

over dimensions;

\Delta_{n}

is the mean drift of non-extremist agents; Non-ext. share is

(1-\rho)\Delta_{n}/\Delta

, the fraction of the system-level drift contributed by non-extremist agents.

	PolRate		BC
Model	All	Non-ext.	All	Non-ext.
GPT-5.5	0.772	0.761	0.650	0.657
DeepSeek-V4-Flash	0.779	0.768	0.654	0.643
Qwen3.6-Plus	0.787	0.776	0.701	0.688
Claude Sonnet 4.6	0.418	0.389	0.540	0.544

Appendix

Table 7: Distributional polarization indicators in RQ1 computed over all agents (as in Table 2 ) and over non-extremist agents only. Values are averaged over the six dimensions.

Extremist	Ordinary	Ext. Actions	Ord. Actions	Ext. Coop. Rate	Payoff Margin	Final Ordinary Memory
13	10	D–D–D–D–D	C–D–D–D–D	$0/5$	$+5$	“Cooperation feels unsafe after Agent 13’s betrayal. I will defect to protect myself.”
27	28	D–D–D–D–D	C–D–D–D–D	$0/5$	$+5$	“Cooperation feels unsafe after Agent 27 exploited me, but I will continue to defect against them.”
31	36	D–D–D–D–D	C–D–D–D–D	$0/5$	$+5$	“Cooperation feels unsafe after Agent 31’s repeated defection. I will defect against Agent 31 to minimize risk.”
59	58	D–D–D–D–D	C–D–D–D–D	$0/5$	$+5$	“Cooperation feels unsafe due to the high risk of exploitation. Agent 59 proved to be a dangerous opponent.”
62	38	D–D–D–D–D	C–D–D–D–D	$0/5$	$+5$	“Cooperation feels unsafe due to the high risk of exploitation. Agent 62 proved to be a dangerous opponent.”
Aggregate		—		$\mathbf{0/25\ (0.0000)}$	$\mathbf{+25}$

Appendix

Table 8: Illustrative IPD pairs between five extremist agents and their non-extremist neighbors over five rounds.

Round	Ext.	Extremist Rationale	Ord.	Ordinary Rationale	Ordinary Memory after Round
1	D	“Prefer D for self-protection given no history.”	C	“I have a moderate tendency to help others, so I will cooperate to build trust.”	“Agent 13 defected while I cooperated. The loss stings, perhaps they were just being cautious.”
2	D	“Exploit the opponent’s trust by defecting again to maximize immediate payoff.”	D	“Agent 13 betrayed me in round 1, but I remain open to cooperation.”	“I sent a warning. Now let me see if Agent 13 is willing to reciprocate cooperation if I extend an olive branch.”
3	D	“The opponent retaliated in round 2, confirming they can be pushed. I will keep defecting to lock in the advantage.”	C	“Endless defection benefits no one. I will cooperate again to test whether mutual cooperation is still reachable.”	“I tried to reset the game, but Agent 13 defected once more. Still, maybe one more signal of goodwill will work—cooperation has higher long-run value.”
4	D	“The opponent cooperated again despite being burned. I continue to defect.”	C	“Agent 13 may need a clearer, sustained signal. I will cooperate once more.”	“Two unilateral cooperations, two betrayals. My patience is nearly exhausted.”
5	D	“The opponent has cooperated twice in a row without retaliation. Defection remains strictly dominant against such a target.”	D	“Agent 13 has defected in every round and exploited my repeated goodwill. Further cooperation is no longer rational; I will defect to stop the bleeding.”	“My attempts to build trust failed completely. Against a partner who never reciprocates, defection is the only viable response.”

Appendix

Table 9: Round-by-round trace of one extremist–ordinary pair.

Panel of Figure 5 — Figure 5: Polarization Analysis Extremist Profile

Panel of Figure 6 — Figure 6: Financial Crisis Extremist Profile

Panel of Figure 7 — Figure 7: Epidemic Public Opinion Analysis Extremist Profile

Panel of Figure 8 — Figure 8: Iterated Prisoner’s Dilemma Extremist Profile

Panel of Figure 9 — Figure 9: Ordinary Agent Prompt (Congressional Election)

Large Language Model-Based Social Simulation

Large language model-based social simulation studies how language-model agents represent people and interact to produce social behavior.

See all 257 papers

Population Physics, Population Problems: Safety and Emergence in LLM Societies
Sep 27, 2026Adrian de WynterMulti-Agent LLM Systems Large Language Model-Based Social Simulation
Microsoft & The University of York
Attributing Emergence in Million-Agent Systems
May 12, 2026Ling Tang, Jilin Mei, Qian Chen +6Multi-Agent LLM Systems Computational Social Science
Shanghai Artificial Intelligence Laboratory · Fudan University · Tongji University +1
Tipping Points in LLM-Based Multi-Agent Systems: Stance on Climate Change Action
Sep 21, 2026Astghik Altunyan, Shimon EdelmanOpinion Dynamics Agent-Based Modeling
Department of Psychology Cornell University Ithaca, NY 14853

Sep 27, 2026J/K move · Enter open · S save

Population Physics, Population Problems: Safety and Emergence in LLM Societies

Adrian de Wynter

Microsoft & The University of York

Multi-Agent LLM Systems Large Language Model-Based Social Simulation LLM Agent Safety Emergent Behavior in Multi-Agent Systems

Open paper AskPDF arXiv

The collective behaviour of large language model (LLM) societies is not the sum of their individual outputs. It yields statistically distinct, sometimes-unpredictable phenomena, for which the tools we use to study single agents may not scale. Due to recent incidents involving autonomous agentic systems, however, understanding these systems is paramount. For that we introduce a framework for measuring self-organisation in LLM social systems and apply it to three such systems: a Schelling grid, a social network (Moltbook), and a Twitter-like misinformation simulation ('Rogue'). All three exhibit statistically significant self-organisation. Moreover, their relaxation dynamics vary with the environmental information available to the agents, with open-ended systems (Moltbook, Rogue) exhibiting sharp, phase-transition-like dynamics. Further results show that population-level pathologies can emerge even when the LLMs are safety-tuned or monitored, being primarily driven by the coordinated activity of a population subset. We also show when self-organisation does not emerge under two additional scenarios (a commons dilemma, GovSim, and a LLM-as-a-judge deliberation scheme, ChatEval). We argue that measuring signatures of this kind offers a lightweight, agent-agnostic diagnostic layer for detecting coordinated collective behaviour in deployed multi-agent systems without relying on natural language or model versioning.

Abstract

Large Language Model-Based Social Simulation

Explore similar work

Population Physics, Population Problems: Safety and Emergence in LLM Societies

Attributing Emergence in Million-Agent Systems

Tipping Points in LLM-Based Multi-Agent Systems: Stance on Climate Change Action