AI’ll take that: Artificial Intelligence, platform’s pricing and partnerships

Artificial Intelligence (AI) companies need large amounts of data to create and im- prove their softwares. Many online platforms allow external actors to access their data through their Application Programming Interface (API), either freely or at a cost. I study how an AI company’s use of the platform’s data affects the platform’s equilibrium strategy regarding its API’s access. I present a model where users and developers join a monopolistic platform that monetizes user data and sustains computational costs. Users freely access the platform, whereas developers pay a price to use the platform’s API. The AI company’s presence leads to an increase in the API price, reducing user participation. If the platform can partner with the AI company, it only does so if its data monetization is low. Under the partnership, the platform further increases the API price to maximize the AI company’s willingness to pay, leading to an inefficient outcome. The model thus proposes a rationale for recent managerial choices made by major social networks such as X (formerly Twitter) and Reddit, who introduced an API price after the entry of ChatGPT.