AI-enabled companies such as Apple, Google and Amazon are faced with a dilemma regarding their platform development: should they open up some of their components to an ever-growing variety of add-ons to enrich their offerings, collect data, train their AI and innovate, at the risk of experiencing cloning of these same components by competitors (forks) and fragmenting their platform ecosystem, or is it preferable for these companies to limit their degree of openness through restrictive access conditions (via their APIs) to prevent themselves from possible forks, then incurring the risk of reducing the integrative scope of their platform? This paper aims to understand how AI renews the question of this strategic trade-off. It also analyzes the different types of strategy available to platform owners to expand their ecosystems while preserving the key assets that contribute to their growth.