Fundamental Market Design as a Layer of AI-Agent Alignment

Jul 16, 2026

Upload by Ivy

The paper argues that AI-agent alignment in markets is not only an agent property, but also a consequence of the interaction infrastructure where agents operate. In financial markets, this infrastructure is the market core: the rule system governing order entry, interaction, matching, persistence, and stabilization. If the core mechanism enables or rewards undesirable behaviour, higher-level alignment may be insufficient. It proposes formal, transparent modelling of the market core to analyze incentives and design mechanisms.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages6

Reading Time15 min

Abstract

Market mechanisms as alignment infrastructure

Alignment as an interaction-layer property

Layered view of alignment

Formalising the market core

Transparency of core interaction rules

arXiv :2607 .09702v1 [ cs .GT] 21 Jun 2026 Fundamental Market Design as a Layer of AI-Agent Alignment ∗ Omar Inverso 1 , Emilio Tuosto 1 , and Dragiša Žunić 1,2 1 Gran Sasso Science Institute, Italy 2 Institute for Artificial Intelligence of Serbia Abstract This paper argues that AI-agent alignment in markets should not be understood only as a property of agents, but also as a property of the interaction infrastructure in which agents act. In financial markets, this infrastructure is the market core: the rule system that determines how orders enter, interact, match, persist, and stabilize. If this fundamental interaction layer allows or rewards undesired behaviour, then higher-level alignment of agents may be insufficient. We propose to view fundamental market design as a layer of AI-agent alignment. Alongside the important work of computational economics in modelling agents, strategies, and learning, we focus on a complementary but more fundamental layer: the formal modelling of the market core itself. Market design, especially at the level of the core mechanism, can benefit from a rigour characteristic of theoretical computer science. This gives a transparentbox model of the market, whose core properties can be formally specified and reasoned about. It also lets us treat the trading venue not as a static order book, but as a computational process combining resident orders with incoming order flow, and ask which computational model, perhaps yet unknown, naturally lies at its core. This perspective is especially relevant for markets populated by adaptive or AI agents. Such agents may learn what the mechanism rewards, including speed, delay, liquidity provision, or manipulation. These behaviours are not only properties of individual agents, but may emerge from the agent–mechanism system. We therefore argue that transparent formal models of market cores can support incentive-oriented analysis and the design of mechanisms in which desirable behaviours are structurally favoured and undesirable behaviours are harder to sustain. 1 Market mechanisms as alignment infrastructure This paper proposes a formal symbolic view of core market mechanisms as a foundation for studying AI agents as market participants. Financial markets involve heterogeneous actors – regular traders, liquidity providers, high-frequency firms, institutional investors, and algorithmic or AI-based systems – but all of them interact through the same rule-governed infrastructure: the trading venue itself. Like a game of chess, a market mechanism defines admissible actions, interaction rules, priority, and outcomes. For this reason, the behaviour of market participants cannot be understood only by modelling their strategies or objectives; it also depends on the formal structure of the mechanism through which they interact. The position of this paper is that alignment is not only a property of agents, but also a property of the interaction infrastructure in which agents act. In financial markets, this infrastructure is the market core: the rule system that determines which orders may interact, ∗ Accepted at the EC’26 Workshop on Incentive-Based AI Alignment, co-located with the 27th ACM Conference on Economics and Computation, Rome, Italy, 2026 . Authors’ email addresses: [omar.inverso@gssi.it](omar.inverso@gssi.it), [emilio.tuosto@gssi.it](emilio.tuosto@gssi.it), [dragisa.zunic@gssi.it](dragisa.zunic@gssi.it). which interactions have priority, how matching is performed, what persists, and when the market has reached a stable state. This perspective is especially relevant for incentive-based AI alignment, because the market core can make some behaviours natural, profitable, and stable, while making others fragile, costly, or impossible. In this sense, market design is an integral part of the alignment environment for AI agents: the core that guides their interaction can make desired behaviours easier to obtain, and undesired behaviours hard

FAQs

为什么作者认为AI-agent alignment不能只看代理本身？

因为代理与市场的互动遵循由市场核心机制决定的规则。核心机制如何允许、匹配、持久化并稳定订单，会直接塑造哪些行为会变得自然、获利或稳定。

文中所说的“市场核心（market core）”具体指什么？

指交易场所的规则系统：决定订单如何进入、如何相互作用、如何匹配、何时与如何持久存在，以及何时达到稳定状态的机制。

作者建议如何用来支持激励导向的对齐分析与机制设计？

通过对市场核心进行透明的形式化建模，使核心性质可被明确指定并进行推理，从而分析激励结构如何让理想行为更容易维持、让不理想行为更难持续。

Jul 16, 2026

Upload by Ivy