Following the integration of ChatGPT into Apple Intelligence, OpenAI discovered that the outcomes fell short of expectations, as evidenced by lower-than-anticipated figures for both new weekly active users and subscription conversion rates. By the summer of 2025, internal assessments within OpenAI had concluded that the integration was ineffective. A key factor contributing to this underperformance was that ChatGPT was not enabled by default in Apple Intelligence; instead, Apple retained control over the access points for AI functionalities, the mechanisms for invoking these features, and the selection of subsequent models. This arrangement significantly curtailed ChatGPT's actual usage. Subsequently, Apple opted to utilize Google Gemini as the foundational model for the new iteration of Siri AI, thereby further consolidating its grip on the core entry points of AI services.