Google Unveils Private AI Compute As Next Step In Delivering Secure and Intelligent AI

In Brief

Google has launched Private AI Compute, a cloud-based platform that combines Gemini models with hardware-secured isolation to deliver advanced AI capabilities while keeping user data fully private.

Technology company Google has introduced Private AI Compute, a cloud-based AI platform designed to process data using Gemini models within a hardware-secured environment, ensuring that user information remains private.

The platform extends Google’s decades-long work in privacy-enhancing technologies (PETs) to the cloud, providing the computational power needed for advanced AI while maintaining the security standards typically associated with on-device processing. Private AI Compute enables faster, more intelligent AI responses, offering personalized suggestions and performing tasks while keeping all data inaccessible to external parties, including Google.

As AI systems evolve from basic task execution to proactive and personalized assistance, they require computational capabilities beyond what most devices can handle locally. Private AI Compute addresses this need by delivering cloud-level performance while keeping user data isolated within a secure environment.

Built on Google’s Secure AI Framework and guided by AI and Privacy Principles, the platform creates a fortified space where personal information and unique insights are protected with additional layers of privacy and security. Private AI Compute runs on an integrated Google tech stack with custom Tensor Processing Units (TPUs) and Titanium Intelligence Enclaves (TIE), combining high-performance cloud computing with robust privacy safeguards.

Through encryption and remote attestation, devices connect to this sealed cloud environment, allowing Gemini models to process sensitive information securely. This setup ensures that all data processed within Private AI Compute remains under the exclusive control of the user, with no access granted to Google or any other party.

Private AI Compute To Enhance On-Device AI While Maintaining Strong Privacy Protections

Private AI Compute allows on-device features to operate with enhanced capabilities while maintaining strong privacy protections. Leveraging this technology, Magic Cue delivers more timely and relevant suggestions on Pixel 10 devices, and the Recorder app on Pixel can now generate summaries of transcriptions across a broader set of languages.

This marks the beginning of a new phase for AI functionality, as Private AI Compute combines on-device processing with advanced cloud models to handle sensitive tasks securely. Further developments and technical insights into how Private AI Compute enhances AI privacy will be shared in upcoming updates.