Data privacy and security will remain essential for Apple’s AI

Apple will relaunch the Apple Intelligence and Siri platforms with new Apple Foundation models. Despite Google’s involvement, Apple will maintain its stance on privacy.

When Apple Intelligence was revealed at WWDC 2024, Apple had a hybrid system that would ultimately fail. Delays ensued and it appears the long wait is over for Apple’s true AI strategy to emerge.

According to the To light up newsletter, Apple won’t compromise on privacy with its new AI efforts. Although the report is tinged with assumptions and conjectures about what’s to come, it paints a pretty clear picture.

Apple will not compromise on privacy for better artificial intelligence.

There are no new details on Apple’s AI efforts. It repeats everything we know about the upcoming strategy and paints a picture of losses, shortcomings and desperation on Apple’s part.

Of course, I don’t see it the same way.

Apple’s place in the AI ​​race

The AI ​​industry has outpaced Apple with increasingly powerful models capable of performing seemingly astonishing tasks. The demos have always been something spectacular, like those in science fiction, but real-world usage has been something a little more mundane.

People are unhappy that their sacrifice of the world’s knowledge and data has achieved very little. Some AI is great and speeds up human workflows, but the cost to our financial markets, component availability, and environment has been incredible.

Apple missed the hype cycle around AI, but still thrived. It continues to post record quarterly results without any significant updates to its AI systems, contradicting the scams being sold to investors.

Apple doesn’t need AI, but AI needs Apple.

With ChatGPT on track to become cash poor by 2028 without an influx of cash and the general public growing increasingly angry with AI companies, Apple’s position couldn’t be stronger. It is not a matter of despair and failure, but success due to patience.

I’m not saying Apple wouldn’t have been happy to see its initial launch go more smoothly. I’m also not saying that Apple wouldn’t have released improved AI sooner if it could have done so.

It all seems like a happy accident. But instead of feeling sorry for itself, Apple is doing what it does best.

Apple is poised to make a late move into the industry with a solution that truly meets people where they are.

Apple’s stance on privacy will be maintained

Expect WWDC 2026 to reveal much of what Apple hopes to accomplish over the next year with AI. However, it won’t reveal everything, like explicit details about Google Gemini usage.

Apple does not replace its models with Gemini

Apple has partnered with Google to get a version of Gemini that can run on Private Cloud Compute servers. No, it does not replace the Apple Foundation models with Gemini, no matter how often they are repeated, but it is used to distill knowledge and train.

There are also rumors that Apple is leasing AI computing space from Google, which is likely due to the state of the market. However, users don’t have to worry about Apple sending data to Google’s servers.

Regardless of which servers and GPU clusters Apple uses, they will perform no differently than Private Cloud Compute with data privacy protections in place. This is no different than Apple renting data servers from Google or Amazon for iCloud.

These companies do not have access to the data. Period.

Apple’s AI strategy

Either way, the new Apple Foundation models will form the backbone of Apple’s new AI strategy. They will run both on the device and in Private Cloud Compute to analyze data and perform tasks on behalf of the user.

In-device AI that stays out of the way

Most users will likely interact with Apple’s AI systems and Siri using these basic models and nothing else. This will be the default, and after Gemini training, it will likely be more than sufficient for the features Apple will unveil at WWDC.

Anyone who wants to ignore AI on Apple platforms will be able to do so.

For those who want other options, Apple offers developers an API. The relationship with OpenAI is fraying and Apple will likely move them out of their privileged position with iOS 27.

Instead, apps like ChatGPT, Gemini or Claude will be able to be installed from the App Store and become endpoints for Apple Foundation templates. This means that whether you invoke Siri or general Apple Intelligence programs, you can send data to a third-party AI for analysis and execution.

Such integrations will maintain user privacy through the use of the API. Apple will likely establish that developers must adhere to strict privacy rules to access the API or risk having access removed.

If Apple doesn’t go this route, Apple will at least warn users about the privacy risks of using third-party models.

Apple has spent over a decade telling users their iPhone is private and secure

The end result is a new set of (hopefully) capable Apple Foundation models powering every AI interaction on iPhone with privacy and security intact. Users will also be able to exploit their favorite AI models, undoubtedly more efficient, according to their needs or desire.

Apple won’t need to be the best in AI. Instead, it will have a sufficiently strong core offering with the ability to support external AI models as part of an expansion of its ecosystem.

Of course, Apple is late in this move, but it seems odd to view this as some sort of desperate move. This is Apple doing what Apple does best, and it’s disrupting an established market with a better business model focused on user needs rather than profits and scams.

Apple will own the AI ​​ecosystem by hosting each model on its powerful hardware while offering good enough models on the device.

WWDC 2026 will begin on June 8 with a keynote address. Expect this to be an AI-focused event, given the amount that will need to be covered in the space.