Apple AI: Privacy is the make-or-break factor.

While Apple may not boast the most cutting-edge artificial intelligence, it is poised to offer what could be the industry's most private AI experience.

As anticipated, the recent WWDC keynote largely centered on AI. Apple strategically reframed its later entry into the AI landscape as a deliberate commitment to "doing things right," with a strong emphasis on unparalleled privacy. This positioning serves as a compelling selling point, though its long-term efficacy remains to be seen.

The newly unveiled Apple Intelligence features and the enhanced Siri AI are engineered for seamless integration across iPhone, iPad, Mac, Apple Watch, and Vision Pro. Highlights include a dedicated Siri AI application featuring a chatbot reminiscent of ChatGPT, advanced AI-powered camera and photo editing capabilities, and foundational agentic functionalities enabling Siri AI to interact with other applications and software on Apple devices.

Regardless of the device used, Apple states that AI query processing will be consistent: handled on-device whenever feasible, and via its secure Private Cloud Compute system when not. Apple assures users that their data will not be stored, will only be utilized to fulfill their specific requests, and will be inaccessible to Apple or any third parties. Conversation logs within the new Siri AI app will be retained exclusively on-device and secured within end-to-end encrypted iCloud accounts.

This architectural approach, particularly Private Cloud Compute, was first introduced in 2024 alongside the initial Apple Intelligence launch. However, two years later, two critical developments have emerged. Firstly, Apple undeniably lags behind most competitors in AI, even after the recent announcements, making its privacy-centric pitch more vital than ever for differentiation. Secondly, this very lag has necessitated partnerships, with Apple now collaborating with Google and Nvidia to power its AI operations.

Consequently, Apple's new cloud AI models are no longer entirely in-house but are built upon Google Gemini. Furthermore, Private Cloud Compute has expanded beyond Apple's proprietary data centers, now operating on Google Cloud systems that leverage Nvidia GPUs, Intel CPUs, and Google Titan chips.

This represents a significant shift. When Private Cloud Compute was initially announced, Apple underscored its design for Apple silicon, emphasizing a hardened supply chain with extensive security scans and validation checks for every server. However, Apple cannot exert the same control over Google, Intel, and Nvidia’s supply chains. Instead, Apple now maintains a "cryptographically verifiable, append-only ledger" for all Google Cloud hardware utilized by Private Cloud Compute and "retains complete control" over the software. Apple claims this hybrid system maintains the same "extraordinary security and privacy properties," though skeptics might contend that the extended supply chain introduces new vulnerabilities.

Nevertheless, Apple can credibly assert that its AI strategy prioritizes privacy to a degree largely absent among other AI giants. This point was explicitly made by SVP of software engineering Craig Federighi during the keynote: “Some appear to be racing forward, seemingly pursuing AI for the sake of AI, without clear regard for the people, all of us, that it’s ultimately meant to serve.” He clarified that Apple Intelligence has been meticulously designed "with privacy in mind at every step."

While comprehensive, Apple's privacy solutions are not entirely unique. Last year, Google introduced Private AI Compute, described at the time as "virtually identical" to Private Cloud Compute, running on a "seamless Google stack" powered by its in-house TPUs. The key distinction lies in their application: Apple deploys its private cloud computing for every AI query that cannot be processed on-device, whereas Google has been less specific about when it utilizes Private AI Compute. Google confirms its use for features like Magic Cue and the AI-powered Recorder app on Pixel phones but has not clarified if it extends to every Gemini query from Google phones, or if similar protections exist for Gemini on other platforms.

In contrast, Google explicitly states that Gemini collects a vast amount of data by default, including submitted prompts, shared files, and recordings of spoken conversations. It also gathers data on the content Gemini generates, the tasks it performs, and information from and about the apps, browsers, and devices used with it. Gemini chat history is typically stored for 18 months before deletion, though users can reduce this period to as little as 72 hours.

Other leading AI companies exhibit similar practices. OpenAI reports collecting user prompts and uploaded content to ChatGPT, along with various location and device data. Chats are used for training data by default, though this can be disabled. Anthropic collects comparable data when users interact with Claude, noting that it deletes audio recordings of spoken prompts while retaining transcripts. Claude also defaults to using user data for model training, stating it is kept in a "de-identified" form for up to five years.

Conversely, Apple's Intelligence privacy policy, last updated in 2025 prior to the recent announcements, specifies that Apple collects only "limited information" regarding Private Cloud Compute requests, such as their size and completion time, crucially excluding any information about the content of the request or its outcome. Apple benefits from not needing to collect user data for model training, as Google has already undertaken this using its own users' data. While Apple's partnership with Google might provoke concerns about privacy maintenance, it may also be the precise mechanism enabling Apple to deliver functional AI without compromising user privacy.

Ultimately, it is Apple's commitment to significantly more limited default data collection that distinguishes its privacy promise. Even if the expanded Private Cloud Compute proves to be less secure than its initial iteration, Apple's claim of gathering substantially less data than any AI alternative is compelling. For many Apple users, this could make Apple Intelligence well worth the wait, whether that be a few months or even years.