Apple says its AI is still private, even when it's running on Google's servers

Apple has announced that its upgraded Siri AI will utilize Google's Gemini language models and run on Nvidia hardware within Google's servers. Despite this shift to third-party infrastructure, Apple maintains its strong privacy promises, similar to when its AI models operated exclusively on user devices or Apple-controlled servers. This move addresses the limitations of on-device processing for more sophisticated AI functions.

Apple has historically championed user privacy, encrypting cloud services and processing data locally whenever possible to minimize data exposure. However, the advanced capabilities required for the new Siri AI necessitated a move beyond its own hardware limitations. The company's Private Cloud Compute system, previously reliant on Apple's servers, has been adapted to function securely on third-party hardware.

To ensure privacy with external servers, Apple is implementing a new iteration of Private Cloud Compute. This system leverages Nvidia's Confidential Computing, Intel's Trust Domain Extensions, and Google's Titan security chip, creating multiple layers of protection. Furthermore, Apple will maintain a cryptographically verifiable ledger of all Google Cloud hardware in the PCC fleet, and devices will only trust Apple-signed software on these servers.

Important decisions, such as model selection and data access for applications, are managed by an on-device "System Orchestrator." This feature ensures that only the necessary data is sent off-device for processing, and all records of that data are immediately vaporized after a query is answered. This transient approach guarantees that no personal data is stored in the cloud.

The new Siri AI and other Apple Intelligence features are set to launch with the upcoming iOS 27, iPadOS 27, macOS 27 Golden Gate, and other operating system releases this fall. Beta versions are currently available to developers, with a more stable public beta expected in July.