Apple confirms Siri AI uses Google's Gemini models while maintaining privacy assurances

Apple announced at its Worldwide Developers Conference that its upgraded Siri, referred to as "Siri AI," will utilize Google's Gemini language models and run on Nvidia hardware located in Google servers. Despite this change, Apple reiterated its commitment to user privacy, which has been a longstanding feature of its platforms. The company emphasizes that its cloud services employ encryption to protect user data and that it has historically favored on-device processing to minimize data transmission. However, the limitations of Apple's hardware have necessitated this shift, as local processing capabilities for language and reasoning models on devices like iPhones and Macs are restricted. Apple's Private Cloud Compute system offered a partial solution but still relied on its own server infrastructure. To fully support Siri AI, Apple would need to invest in significant data center expansion, which it has not yet pursued.