On the Antitrust Implications of Embedding Generative AI in Core Platform Services

Abstract

Since the rise of ChatGPT, digital gatekeepers have integrated generative AI into their core services, potentially exacerbating harms from anti-competitive behaviors. Generative AI enables these platforms to repurpose content from business users to meet end-user demand directly, reducing the value provided by these business users. Consequently, platforms shift from intermediaries to direct suppliers, centralizing information and diminishing the need for users to leave their ecosystem. This centralization, combined with personalized content creation and opaque advertising practices, could lead to unprecedented levels of user manipulation. Business users, in turn, may face "exploitative platform discrimination," creating inescapable prisoner’s dilemmas that allow platforms to extract excessive rents. 

This article outlines the potential harms from embedding generative AI into core platform services and proposes several measures to mitigate them. These measures include rebalancing economic control over creative outputs, severing anti-competitive links between content dissemination and creation, and exposing ‘preferred partnerships’ that undermine legitimate industry responses. By addressing these issues, we can prevent generative AI from consuming the web that enabled its rise and ensure it causes more good than harm.

JEL Classification: K21, L12, L41, L86, M37, O31