生成式AI轴辐协议对反垄断法意思联络理论的挑战与重构

The Challenge and Reconstruction of the Antitrust Law’s Meaningful Communication Theory by Generative AI Hub-and-Spoke Agreements

摘要: 生成式人工智能(以下简称“生成式AI”)作为定价中枢，催生了无需人类明示合意即可涌现协同定价的新型轴辐协议。传统反垄断法“意思联络”要件在归责筛选、证据门槛与罪责匹配上遭遇三重失灵。本文以“智定价”虚拟案例为线索，解构人机混合决策的三种法律形态，借鉴德国“组织过错”与“保证人义务”理论、美国反托拉斯法“便利化行为”规制经验及产品责任法“设计缺陷”归责逻辑，提出以“共同风险体”和“算法代理”为核心的双层归责理论，将意思联络从主观合意转向客观风险创设与控制。在制度层面，通过设计共同风险体成立标准、三级合规制度、相适应的责任配置，为反垄断法回应自主智能系统奠定法理基础。

Abstract: Generative artificial intelligence (hereinafter referred to as “generative AI”), as a pricing hub, has given rise to a new type of hub-and-spoke agreement that enables coordinated pricing without requiring explicit human consent. The traditional antitrust law requirement of “mutual intent” has encountered triple failures in liability screening, evidentiary thresholds, and culpability matching. Using the “Smart Pricing” hypothetical case as a thread, this paper deconstructs three legal forms of human-machine hybrid decision-making. Drawing on Germany’s theories of “organizational fault” and “guarantor obligation,” the U.S. antitrust law’s regulation of “facilitating arrangements,” and the product liability law’s “design defect” liability logic, it proposes a two-tier liability theory centered on the “common risk body” and “algorithmic agent.” This shifts the focus of mutual intent from subjective consent to the creation and control of objective risks. At the institutional level, by establishing standards for the formation of a common risk body, implementing a three-tier compliance system, and configuring proportional liability, it lays a jurisprudential foundation for antitrust law to address autonomous intelligent systems.