Apple and OpenAI Partnership: What about user privacy?

At WWDC24, Apple officially entered the generative AI and large language models (LLMs) era and announced a partnership with OpenAI, raising some questions about “Privacy”.

Since the beginning, Apple kept its software ecosystem closed in the name of user data safety and privacy. It drew industry-wide criticism for related practices. However, the EU recently forced Apple to make the majority of these practices available to third-party developers.

This year, the Cupertino-based company has introduced new AI features based on on-device and cloud-based language models. The company also announced the inclusion of ChatGPT capability throughout iOS, iPadOS, and macOS.

Apple says the new AI features run entirely on devices. Requests that require more processing power will be sent to Private Cloud Compute (PCC) to provide better performance and keep the user data submission private. PCC is built with custom Apple silicon, an operations system designed for privacy and security architecture.

That summarizes Apple’s part on AI. Now let’s move on to OpenAI, which will integrate ChatGPT access for iPhones, iPads, and Mac devices. The keynote detail revealed that the users can ask to identify text or documents with ChatGPT via Siri.

However, users will be asked whether to send these queries to ChatGPT or process them via Siri alone. If allowed, the customer will see an answer from ChatGPT right on the screen inside Siri.

OpenAI said its ChatGPT will not record IP addresses or store requests for training. However, these situations may not be the same in the case of connecting a ChatGPT account to these operating systems.

Apple and OpenAI have explained the theory about their privacy features but the picture remained blurred. The ChatGPT integration into operating systems and private computing may bring more confusion to these theoretical answers than a clarification.