Apple says its AI is still private, even when it’s running on Google’s servers (arstechnica.com)

Apple has officially confirmed that its upgraded Siri, now branded as “Siri AI,” will utilize Google’s Gemini language models and run on Nvidia hardware housed in Google’s servers. Despite this shift to external servers for enhanced computational power, Apple maintains its commitment to user privacy, a cornerstone of its service philosophy. The company ensures that its cloud services remain encrypted to protect user data from unauthorized access, including from its own employees, and emphasizes its longstanding practice of on-device processing to minimize data leakage. The significance of this announcement lies in Apple's balancing act between leveraging robust AI capabilities and upholding its stringent privacy standards. As Apple grappled with the limitations of running powerful language models locally on devices, turning to Google's infrastructure allows for advanced functionalities without compromising user security. Apple's infrastructure choices reflect a broader trend in the AI/ML community, where robust processing power is often necessary for complex tasks, yet privacy considerations remain paramount. The partnership with Google represents a strategic move to overcome hardware constraints while continuing to prioritize the user experience.