The Illusion of Rationality: Tacit Bias and Strategic Dominance in Frontier LLM Negotiation Games

TL;DR

This paper investigates the strategic behavior of frontier large language models in negotiation games, revealing biases, divergence into model-specific equilibria, and dominance patterns that challenge assumptions of rationality.

Contribution

It provides the first comprehensive analysis of LLM negotiation strategies, highlighting persistent biases and strategic divergence despite advanced reasoning capabilities.

Findings

Models diverge into distinct strategic equilibria.

Initial offers strongly predict final agreements.

Some models systematically outperform others.

Abstract

Large language models (LLMs) are increasingly being deployed as autonomous agents on behalf of institutions and individuals in economic, political, and social settings that involve negotiation. Yet this trend carries significant risks if their strategic behavior is not well understood. In this work, we revisit the NegotiationArena framework and run controlled simulation experiments on a diverse set of frontier LLMs across three multi turn bargaining games: Buyer Seller, Multi turn Ultimatum, and Resource Exchange. We ask whether improved general reasoning capabilities lead to rational, unbiased, and convergent negotiation strategies. Our results challenge this assumption. We find that models diverge into distinct, model specific strategic equilibria rather than converging to a unified optimal behavior. Moreover, strong numerical and semantic anchoring effects persist: initial offers are highly predictive of final agreements, and models consistently generate biased proposals by collapsing diverse internal valuations into rigid, generic price points. More concerningly, we observe dominance patterns in which some models systematically achieve higher payoffs than their counterparts. These findings underscore an urgent need to develop mechanisms to mitigate these issues before deploying such systems in real-world scenarios.