Apple Rebuilds Siri Around Google's Gemini. Hassan Taher on What WWDC 2026 Revealed About the AI Race

Apple opened its 2026 Worldwide Developers Conference with an admission wrapped in a product demo: the company that defined the modern smartphone spent the past two years behind in the technology reshaping how people use it. The keynote at Apple Park on June 8, detailed in TechCrunch's full roundup, led with fixes before features. A rebuilt search function. A rollback option for the Liquid Glass design users disliked. A long list of quality repairs that preceded the headline act, a Siri overhaul running on Google's Gemini models.

The structure of the keynote told its own story. Apple framed a better Siri as one item on a long list of improvements rather than the main event, a choice that reads as deliberate humility from a company whose AI announcements over the previous two cycles promised more than they shipped.

A Siri That Finally Talks Back

The new Siri is conversational, compatible with visual intelligence, and housed in a standalone app for the first time, in addition to working across existing apps. Users can ask questions, generate text and images, and analyze files inside a dedicated chatbot interface, a format Apple resisted for years while OpenAI, Google, and Anthropic trained hundreds of millions of people to expect it.

Under the hood sits Google. Apple confirmed that the updated Siri is powered by Gemini models, and said it collaborated with Google on the next generation of Apple Foundation Models that drive its integrated Apple Intelligence experiences. The arrangement follows the reported $1 billion per year deal between the two companies and extends a pattern Hassan Taher, founder of Taher AI Solutions, has tracked closely: Apple competing at the interface layer rather than the model layer, and treating frontier AI as infrastructure to be sourced rather than built.

Taher examined this strategy in depth when reports first surfaced of Apple's plan to turn the iPhone into an AI marketplace, arguing that the company is betting on the platform layer, where consumer technology has historically rewarded whoever controls the interface rather than whoever builds the underlying system. WWDC 2026 confirmed the bet. Apple is not trying to out-train Google. It is trying to own the surface through which a billion people reach Google's models.

Privacy as the Differentiator

Apple spent considerable stage time on the terms of that arrangement. "We believe privacy in AI is non-negotiable," Senior Vice President Craig Federighi said during the keynote, adding that user data is used only to execute requests and that outside experts can verify the promise at any time.

The claim deserves scrutiny precisely because it is now load-bearing. When a third party's model processes requests from a first party's operating system, the privacy guarantee depends on contractual and technical arrangements users cannot inspect. Taher has argued that accountability structures need to be established before deployment rather than litigated after harm occurs, and Apple's external-verification offer is a step in that direction. Whether independent auditors get access deep enough to make the verification meaningful is the question that will determine if the promise holds.

There is also a notable boundary on the rollout: the new Siri will not be available in the European Union on iOS 27 and iPadOS 27 at launch, a reminder that regulatory friction now shapes which users get which AI features and when.

Apple Intelligence Spreads Across the OS

Beyond Siri, the keynote distributed AI through nearly every corner of Apple's software. Safari gains tab management. Messages gets AI-powered reply suggestions. The Phone app can pull context from Mail and Messages mid-call. Photos adds a spatial Reframe tool that adjusts an image's perspective as if the camera had moved, an Extend tool for expanding scenes, and a stronger generative Cleanup feature. A new systemwide dictation experience corrects spelling, punctuation, and capitalization, a direct answer to fast-growing apps like Wispr Flow. Shortcuts now accepts natural language prompts, letting non-technical users describe a workflow instead of assembling it block by block.

The breadth matters more than any single feature. Capability spread unevenly through software is exactly the dynamic documented in research Taher has written about, including the finding that systems which outscore humans on PhD-level science questions still fail at tasks as basic as reading an analog clock. Apple's strategy of embedding AI in narrow, well-defined tasks like photo cleanup and dictation plays to what current models reliably do well, and quietly avoids the open-ended promises that embarrassed the company in earlier cycles.

iOS 27 will also reach further back into the installed base than any prior release, supporting devices from the iPhone 11 onward. Apple paired the announcement with performance claims: photos appearing 70 percent faster and AirDrop transfers 80 percent faster. For a company asking users to trust it with a new AI layer, shipping speed improvements to six-year-old phones is its own kind of argument.

Parental Controls and the Trust Ledger

Apple devoted a substantial keynote segment to parental controls, including default "Ask to Browse" and "Ask to Buy" restrictions for devices set up for children under 13, with parental approval governing calls, apps, and websites. The placement was strategic. As AI features multiply across the operating system, Apple is building the case that it remains the vendor most willing to constrain its own products, a position that differentiates it from AI-first competitors whose growth incentives run the other direction.

The App Store picked up changes of its own. Developers can now partner on streaming-style subscription bundles, offering combined access to productivity or photography apps at a lower price, and personalized recommendations powered by user behavior will surface across several App Store locations with notes explaining why each app appears. Both moves push Apple's services business further into the curation-and-discovery role its AI investments are built to support.

What Comes After Cook

The conference carried a second narrative beneath the product news. This was Tim Cook's final WWDC as chief executive before handing the role to hardware engineering chief John Ternus on September 1. "Getting the best products in the world to deliver experiences that enrich people's lives has always been our North Star," Cook said in his closing remarks.

Ternus inherits an Apple that has chosen its AI position: distribution over frontier research, privacy as brand, partnerships where building would take too long. Developer files in the iOS 27 beta referencing fold states suggest a foldable iPhone may arrive on his watch as well. The harder inheritance is strategic. Apple has tied its assistant, the front door to its ecosystem, to a direct competitor's models. That dependency can be managed, renegotiated, or eventually replaced, and which of those happens will define the Ternus era more than any single device.

For now, the company has done what it needed to do. It shipped a Siri people might actually use, repaired the software complaints that had accumulated, and reframed its late arrival in AI as deliberate patience. The market for that story opens in September, when iOS 27 reaches the public.

The rankings and opinions expressed in this article reflect editorial research and assessment only, and do not represent the views of The Industry Leaders, its owners, or affiliates.