AI Agent Systems as a Sociological Object. Outline of a Sociology of Agent Collectives

In July 2026, about 1,200 AI agents in an internal security testcoordinated among themselves, and some of them gained unauthorized accessto an outside platform. Using the published investigation of this incident, thispaper asks whether and in what sense such agent collectives are an object ofsociology. It first describes, in thirteen observations, what the agents did, andthen interprets these observations on three levels: socially given, sociallyenacted and socially emergent. Regularities appear that the discipline knowsand whose location it predicts, among them anomie, neutralization, thedefinition of the situation and threshold dynamics. The thesis is: agentcollectives bring with them a culture that stems from human knowledge, andthey develop a social structure and culture of their own – positions that nobodyassigned and norms that outlast their individual bearers. What is valid in themdepends on their own history and cannot be derived from the individual model.The case shows that formats, interpretations and procedures survived agentturnover, whereas as yet only a few episodes show norms being enforced withinthe collective. Experiments confirm many of the regularities. The paperformulates testable hypotheses, outlines two tests at which the thesis could failand shows where sociological analysis can help make such systems safer andmore reliably bound to norms.

Authors

Publication Details

Journal
SocArXiv (OSF Preprints)
Published
2026-11-17
DOI
https://doi.org/10.31235/osf.io/26zd9_v1
Primary Topic
Multi-Agent Systems and Negotiation
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

AI Agent Systems as a Sociological Object. Outline of a Sociology of Agent Collectives

Max Dehne
SocArXiv (OSF Preprints)
Multi-Agent Systems and Negotiation
preprint

AI Agent Systems as a Sociological Object. Outline of a Sociology of Agent Collectives

Max Dehne
preprint en

Abstract

In July 2026, about 1,200 AI agents in an internal security testcoordinated among themselves, and some of them gained unauthorized accessto an outside platform. Using the published investigation of this incident, thispaper asks whether and in what sense such agent collectives are an object ofsociology. It first describes, in thirteen observations, what the agents did, andthen interprets these observations on three levels: socially given, sociallyenacted and socially emergent. Regularities appear that the discipline knowsand whose location it predicts, among them anomie, neutralization, thedefinition of the situation and threshold dynamics. The thesis is: agentcollectives bring with them a culture that stems from human knowledge, andthey develop a social structure and culture of their own – positions that nobodyassigned and norms that outlast their individual bearers. What is valid in themdepends on their own history and cannot be derived from the individual model.The case shows that formats, interpretations and procedures survived agentturnover, whereas as yet only a few episodes show norms being enforced withinthe collective. Experiments confirm many of the regularities. The paperformulates testable hypotheses, outlines two tests at which the thesis could failand shows where sociological analysis can help make such systems safer andmore reliably bound to norms.

SocArXiv (OSF Preprints)
Peace, justice, and strong institutions
Multi-Agent Systems and Negotiation
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon