In the early hours of June 9 Beijing time, Apple's WWDC 2026 keynote concluded. This was Cook's last WWDC as CEO; he will step down on September 1, with John Ternus, head of hardware engineering, succeeding him.
But compared with personnel changes, after browsing all the published content, I found that three things are directly relevant to enterprise IT leaders—and in some respects, more important than new hardware.
First thing: Siri has become a standalone application, and its core runs on Google Gemini
Apple officially releases Siri AI—not an upgrade, but a complete overhaul. The all-new Siri has a standalone app, cross-app contextual awareness, and visual recognition capabilities. It can understand what you discuss in Messages, find the corresponding email in Mail, and then help you create reminders or notes.
The key detail is hidden beneath the surface: Siri AI's underlying model is powered by Google Gemini. Apple's own Apple Foundation Models were developed in collaboration with Google. Apple's software chief Craig Federighi repeatedly emphasized privacy on stage—"We believe privacy in AI is non-negotiable"—and promised that external experts could verify data usage at any time. But what he did not make clear was: when employees' Siri requests are packaged and sent to Google's cloud model, how many nodes the data passes through along the way.
This is not a technical issue, but a compliance issue. If your company has data export review requirements, then the modified Siri on employees' phones is now a new entry point and needs to be included in the scope of data governance.
The second thing: iOS 27 will support free switching between third-party AI models
Although Apple's official press release did not mention it, TechCrunch and multiple tech media outlets have confirmed: iOS 27 has opened up the ability to switch to third-party AI models. When Siri encounters a question it cannot answer, users can seamlessly switch to ChatGPT, Claude, or other third-party models—without leaving the Siri interface.
This design has a greater impact in enterprise scenarios than on the consumer side. In the past, when employees used third-party AI tools, they at least had to actively open a webpage or application—the IT department at least knew this behavior was happening. Now this switch is built into the system layer, and employees may not even know which model they are using.
This is not to say that this feature is bad. The question is whether an enterprise's data classification, DLP policies, and endpoint management strategies can keep up with this change.
The number of Siri AI features available to mainland Chinese users on iOS 27. Apple officially confirms: Due to regulatory requirements, Siri AI and new Apple Intelligence features will initially be completely unavailable in mainland China.
The third thing: Mainland China is explicitly excluded from Siri AI
Apple was more tactful in its press release—“meeting regulatory requirements.” But what does that mean? It means that for Apple device users in mainland China, the core AI features announced at this WWDC will launch with the devices, but with reduced functionality.
This has implications for enterprise IT at two levels.
The first layer is straightforward: if your company provides iPhones to employees, you can temporarily set aside security policies related to Siri AI—because it simply cannot be used. However, DMA compliance restrictions in the European market also introduce uncertainty—Apple has restricted Siri AI on iOS/iPadOS in the EU, but it is available on Mac. If you have employees in European branches, policies should be managed separately by device type.
The second layer is more subtle: the fact that something is unavailable in mainland China does not mean employees of mainland Chinese companies will not find ways to use it. VPNs, switching Apple ID regions, sideloading—these practices already emerged during the previous round of ChatGPT restrictions. Apple's direct integration of AI features into the system layer may instead strengthen the motivation to “bypass” usage.
There are also some details worth noting
Practical advice for CIOs and IT leaders
If you are an enterprise IT manager, there is no need to hold an emergency meeting about WWDC. However, there are a few things worth completing in the coming weeks:
First, update the endpoint security policy.Siri AI is temporarily unavailable in mainland China, but this gives you a window of time. Before iOS 27 (this fall) is officially released, clarify Siri AI's data flow paths—specifically which data is sent to which servers—and then make a judgment based on your company's data classification and grading standards.
Second, prepare a management plan for “bringing your own AI to work.” Once Apple opens up third-party model switching, employees’ use of AI on corporate devices will become even more invisible. Rather than reacting passively when that time comes, it is better to start drafting a preliminary AI usage policy now—at least clarifying the rules on “what data cannot be entered into AI tools.”
Third, watch whether Apple's promise of “privacy verifiability” can be genuinely implemented.Federighi said that external experts can verify the data processing methods at any time. If this mechanism is truly established, it could become a new model for enterprise AI compliance—but so far, it remains only a promise.
Three major events at WWDC 2026: Siri rebuilt, Gemini arrives, and unavailable in mainland China. For enterprise IT, the first two are technical issues, while the third turns a technical issue into a compliance risk. The window is about three months—enough time.
