Google is entering the privacy-focused AI cloud space with its new Private AI Compute platform. Designed to bring the benefits of on-device AI privacy to cloud processing, this solution offers enhanced AI capabilities while prioritizing data security.
Similar to Apple’s Private Cloud Compute, Google’s offering leverages its Gemini models within a highly secure environment. Data processed through Private AI Compute is isolated and protected using Google’s robust infrastructure, including custom Tensor Processing Units (TPUs) and Titanium Intelligence Enclaves (TIE), providing an extra layer of defense against unauthorized access.
Google emphasizes its commitment to zero access assurance, asserting that even the company itself cannot access the data being processed within Private AI Compute. This commitment is built upon Google’s Secure AI Framework (SAIF), AI Principles, and Privacy Principles, highlighting its dedication to responsible AI development. Private AI Compute aims to strike a balance between the privacy advantages of local AI models and the computational power of cloud-based intelligence, ushering in a new era of secure and performant AI applications.