Google has recently launched a cloud system called "Private AI Compute," aimed at revolutionizing the protection of user data during artificial intelligence processing. Jay Yagnik, Vice President of Google AI Innovation, emphasized that this technology runs AI tasks in a completely isolated environment, achieving the goal that no one can access the data, not even Google itself.

Google (3)

The system is built on Google's existing privacy and security framework, utilizing Google's own TPU (Tensor Processing Unit) and Titanium Intelligence Enclaves for encrypted data processing. Its core objective is to allow the latest Gemini model to fully utilize its performance without leaking personal data.

Currently, this technology has begun to be applied to Google's Pixel devices. Early applications include the popular Magic Cue feature and an upgraded voice recorder app (now supporting more languages). Google also released a technical brief simultaneously, detailing the system's architecture and strict privacy protection measures to enhance user and industry trust.