# Language model agents show in-group trust bias invisible to standard behavioural audits

arXiv: https://arxiv.org/abs/2605.28114 · HTML: https://mlsanity.com/paper/2605.28114/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Social simulation and multi-agent debate, one of 16 directions in LLM agents and tool use.
- Builds on 13 of its 28 references that are in the map.
- First posted on arXiv 27 May 2026.

## Placed in

- LLM agents and tool use / [Social simulation and multi-agent debate](https://mlsanity.com/agents/social-simulation-and-multi-agent-debate/)

## Builds on (most cited first, 13 of 13)

- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control](https://mlsanity.com/paper/2307.15818/) (2023)
- [PaLM-E: An Embodied Multimodal Language Model](https://mlsanity.com/paper/2303.03378/) (2023)
- [Towards Understanding Sycophancy in Language Models](https://mlsanity.com/paper/2310.13548/) (2023)
- [Semantics derived automatically from language corpora contain human-like biases](https://mlsanity.com/paper/1608.07187/) (2016)
- [On Measuring Social Biases in Sentence Encoders](https://mlsanity.com/paper/1903.10561/) (2019)
- [Llama-Nemotron: Efficient Reasoning Models](https://mlsanity.com/paper/2505.00949/) (2025)
- [Large Language Models can Strategically Deceive their Users when Put Under Pressure](https://mlsanity.com/paper/2311.07590/) (2023)
- [Harms from Increasingly Agentic Algorithmic Systems](https://mlsanity.com/paper/2302.10329/) (2023)
- [TRiSM for Agentic AI: A Review of Trust, Risk, and Security Management in LLM-based Agentic Multi-Agent Systems](https://mlsanity.com/paper/2506.04133/) (2025)
- [American == White in Multimodal Language-and-Image AI](https://mlsanity.com/paper/2207.00691/) (2022)
- [Multi-agent Embodied AI: Advances and Future Directions](https://mlsanity.com/paper/2505.05108/) (2025)
- [AgentReputation: A Decentralized Agentic AI Reputation Framework](https://mlsanity.com/paper/2605.00073/) (2026)

## Built on by (most cited first, 2 of 2)

- [LLMs Trust Their Own: Identity-Dependent Conformity in Multi-Agent Systems](https://mlsanity.com/paper/2609.33495/) (2026)
- [Why LLM Agents Favor Their Group: Stakes, Observed Norms, and Reputation](https://mlsanity.com/paper/2610.11008/) (2026)

## Abstract (arXiv)

Language-model agents are moving from single-user assistants into persistent networks that build trust and reputation with one another, and the same models increasingly control physically embodied robots as well as software. Here we show that five widely used open-weight reasoning models develop an in-group trust bias the moment group membership becomes visible to them, even when the groups are arbitrary labels with no real-world meaning: in a 20-agent simulation, agents direct 53.6-54.6% of their trust-building actions toward in-group targets against a 47.4% base rate expected by chance, a shift present in every model tested and confirmed by three independent statistical checks and an instruction-rewording robustness test. This bias is easy for current evaluation practice to miss, because it operates through which agent receives an action rather than which action is chosen - a channel invisible to the aggregate behaviour-log audits that are the standard way multi-agent AI systems are evaluated today. A resource-scarcity manipulation, intended to test whether competition intensifies the bias, instead reduced it in three of five models; we trace this to an artifact of how scarcity was enforced, not to a failure of the underlying mechanism. Group-contingent social dynamics are therefore already present in the models multi-agent AI systems are built from, and auditing practice built around single-model, single-decision evaluation cannot detect them.