Apple prioritizes privacy for 2026 AI

- Apple told reporters on May 17 it will keep privacy and data security central as it rebuilds Siri and Apple Intelligence around Apple Foundation Models. - Apple’s most specific privacy claim remains that Private Cloud Compute data is “not accessible” to Apple and is removed after requests. - Apple is expected to outline more of its AI roadmap at WWDC 2026, with Siri and Apple Intelligence updates centered on Apple Foundation Models.

Apple said its next round of artificial intelligence products will keep privacy and data security at the center as it rebuilds Siri and Apple Intelligence around its own foundation models, according to reporting published on May 17. AppleInsider, citing what it said was Apple’s briefing with reporters and broader reporting around the company’s 2026 AI plans, said Apple would continue to emphasize on-device controls even as it works with outside partners. Apple has not published a new standalone announcement laying out those 2026 plans in detail. Its existing public documentation, however, shows the privacy architecture the company says will underpin Apple Intelligence and any more advanced cloud-based requests. ### What, exactly, is Apple saying stays protected? Apple’s public support and legal disclosures say Apple Intelligence is designed to use on-device processing whenever possible, so requests can be handled without data leaving a user’s device. For more complex tasks, Apple says the system can send only the data relevant to a request to Private Cloud Compute, a server-based system running on Apple silicon. Apple says that data is not stored, is not made accessible to Apple, and is processed only to fulfill the request. (appleinsider.com) Apple Security Research said on June 10, 2024 that Private Cloud Compute was built so personal user data sent to the service “isn’t accessible to anyone other than the user — not even to Apple.” The company also said independent security researchers can inspect the software running on those servers to verify that privacy promise. ### How does that fit with Apple Foundation Models? (support.apple.com) Apple’s machine learning research team said on June 10, 2024 that Apple Intelligence is built on multiple generative models, including an on-device language model with about 3 billion parameters and a larger server-based language model available through Private Cloud Compute. Apple said those models were built for tasks such as rewriting text, summarizing notifications and taking actions across apps. (security.apple.com) Apple also said those models were developed under responsible-AI principles that include protecting privacy through on-device processing and Private Cloud Compute. That matters because the company’s current privacy case for future Siri upgrades rests on the same architecture: do as much as possible on device, and route harder requests through Apple-controlled cloud infrastructure with limited data exposure. That last point is an inference from Apple’s current documentation and the May 17 AppleInsider report, not a new public Apple statement. (machinelearning.apple.com) ### Where do outside partners come in? AppleInsider reported on May 17 that Apple plans to work with outside partners while keeping privacy safeguards and user controls in place. The report said Google’s involvement would not displace Apple’s own privacy posture and described Apple Foundation Models as the core of the company’s relaunch of Siri and Apple Intelligence. (machinelearning.apple.com) Apple’s legal privacy page does not name any future 2026 partner arrangements, but it does describe the operating rules Apple says govern Apple Intelligence requests: only necessary data is sent, the content is not retained, and Apple collects only limited, non-identifiable metadata such as request size and response time. ### What can users check for themselves? Apple Support says users can generate an Apple Intelligence Report on iPhone from Settings > Privacy & Security to review requests sent to Private Cloud Compute over the last 15 minutes or the last seven days. (appleinsider.com) Apple says that reporting is part of its transparency approach around how cloud-based Apple Intelligence requests are handled. (apple.com) Apple Security Research said researchers can also inspect the software running on Private Cloud Compute servers. That verification mechanism has become a central part of Apple’s public argument that cloud AI processing can be audited rather than accepted on trust alone. ### When will Apple put fuller details on the record? WWDC 2026 is the next obvious venue for Apple to spell out how Siri and Apple Intelligence change under its 2026 roadmap, and AppleInsider said May 17 that the conference is expected to reveal more of what Apple plans to do with AI over the following year. (support.apple.com) Until then, the clearest on-the-record details remain Apple’s existing privacy, support and security documents, which describe the on-device and Private Cloud Compute system Apple says will govern future AI features as well. (security.apple.com) (appleinsider.com)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.