Google has announced a new cloud platform that makes it possible to deliver advanced artificial intelligence (AI) functions while protecting the privacy of user data. This system, called Private AI Compute, is designed to securely process AI requests that exceed the processing capacity of the devices in the cloud. This structure is very similar to the Private Cloud Compute system that Apple introduced before. Google’s new approach aims to establish the balance between users’ increasing privacy expectations and the high computing power required by today’s artificial intelligence tools.
Until now, Google has been running many AI functions directly on the device, such as translation, voice summaries, and chat assistants. In this way, user data could be processed without leaving the phone, tablet or Chromebook. However, Google states that this approach is not sustainable in the long term. Because more advanced AI models require a capacity beyond the processing power offered by the devices. At this point, Private AI Compute comes into play.
Google combines cloud power with on-device security
According to Google, Private AI Compute can protect the privacy of user data as if it were on the device, even though it works in the cloud environment. While the company emphasizes that this infrastructure offers a “secure and hardened area”, it states that the data belongs only to the user and no one, including Google, can access this information. This approach aims to offer a remarkable balance of both performance and security.
The additional processing power offered by the platform allows AI features to go beyond performing simpler tasks and provide more personalized and contextual recommendations. For example, the Magic Cue tool introduced by Google can provide more relevant suggestions by pulling contextual information from e-mail and calendar applications. Thanks to this feature, which will be used on Pixel 10 series devices, users will be able to provide suggestions that are more compatible with their agenda. In addition, the transcription feature offered by the Recorder application will also support more languages.
Google states that the Private AI Compute platform is just the beginning. This technology is expected to be integrated into more products and services in the future. The company also plans to publish technical documentation as open source and offer a structure open to external audits in order to increase confidence in the security of the system. This approach will allow both developers and security experts to evaluate the system.
In addition, there is a general trend across the industry towards delivering AI services in cloud-based but privacy-focused formats. In addition to Apple, companies such as Samsung and Microsoft are also working on similar solutions. These developments indicate that a more controlled and user-oriented structure will be adopted in the future of AI-based digital services.