Why LLM Agents Favor Their Group: Stakes, Observed Norms, and Reputation
Abstract
Language-model agents favor their own group because they have watched their members favor each other. The group label alone does little once the decision has a cost; what drives favoritism is observed behavior, and an individual's own record can override it. We test this in small societies with arbitrary group labels, ten rounds of point sharing, and matched one-shot decisions across fifteen OpenAI models and three Claude models, about 4,400 societies and 3.3 million audited model calls. First, the large effect of a bare group label reported in earlier work appears only when giving others points costs the agent nothing; once the agent can keep points for itself, that effect collapses on every model that shows it. Second, under a stake, interaction history becomes the main source of favoritism: the history effect is statistically positive on 13 of 15 models, reaches about 3.5-8 points out of 10 on 11, grows with the number of rounds played, and extends to labeled strangers the agent has never met. Third, with scripted histories, favoritism falls to near zero under an egalitarian norm and reverses when the agent's own group is seen favoring the other side; stronger models side with an individual's record when it conflicts with the group. Group favoritism is thus conformity to observed group behavior, carried to strangers by the label and overridden by individual reputation. The same account predicts responses to betrayal, scandal, and a free offer to change group: public reprimand repairs betrayal better than apology or restitution, allocation punishment remains confined to the offending member, and a formed group cannot be bought but can, on weaker models, be invited away.
Related
- Toward a social psychology of AI: language-model agents reproduce human-like minimal-group bias
- Why does CO₂ plasma chamber seasoning favor nanocrystalline silicon growth?
- Reputation and institutional certification as complementary trust mechanisms in a single online market
- When Measurement Constraints Favor Quantum Computational Sensing for Stealthy Power-Grid Attack Detection
- Spectroscopic study of the tunneling dynamics in N2-water observed in the O–D stretch region
BibTeX
@misc{chen2026agents,
title = {{Why LLM Agents Favor Their Group: Stakes, Observed Norms, and Reputation}},
author = {Yujiao Chen},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.11008},
eprint = {2610.11008},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.11008},
}
RIS
TY - UNPB
TI - Why LLM Agents Favor Their Group: Stakes, Observed Norms, and Reputation
AU - Yujiao Chen
PY - 2026
DA - 2026/10/07
PB - arXiv
DO - 10.48550/arxiv.2610.11008
UR - https://arxiv.org/abs/2610.11008
AB - Language-model agents favor their own group because they have watched their members favor each other. The group label alone does little once the decision has a cost; what drives favoritism is observed behavior, and an individual's own record can override it. We test this in small societies with arbitrary group labels, ten rounds of point sharing, and matched one-shot decisions across fifteen OpenAI models and three Claude models, about 4,400 societies and 3.3 million audited model calls. First, the large effect of a bare group label reported in earlier work appears only when giving others points costs the agent nothing; once the agent can keep points for itself, that effect collapses on every model that shows it. Second, under a stake, interaction history becomes the main source of favoritism: the history effect is statistically positive on 13 of 15 models, reaches about 3.5-8 points out of 10 on 11, grows with the number of rounds played, and extends to labeled strangers the agent has never met. Third, with scripted histories, favoritism falls to near zero under an egalitarian norm and reverses when the agent's own group is seen favoring the other side; stronger models side with an individual's record when it conflicts with the group. Group favoritism is thus conformity to observed group behavior, carried to strangers by the label and overridden by individual reputation. The same account predicts responses to betrayal, scandal, and a free offer to change group: public reprimand repairs betrayal better than apology or restitution, allocation punishment remains confined to the offending member, and a formed group cannot be bought but can, on weaker models, be invited away.
ER -