Apple's WWDC keynote on June 8 did something unusual for Apple: it opened with an apology. Before showing anything new, the company's software chief Craig Federighi spent the first stretch of the keynote fixing things that were broken. A controversial design overhaul from last year was dialed back with a new slider. A file-sharing feature that routinely failed was repaired. The phone's built-in search was described as rebuilt from scratch. Only then did Apple turn to Siri.
The structure said everything. Apple is aware that its credibility with users has taken real hits over the past two years, and it led with repairs before making new promises.
The new assistant is officially called Siri AI. It is a full chatbot, not just a voice trigger. A dedicated app stores your conversation history, accepts file and photo uploads, and syncs privately across all your Apple devices. A swipe down from the top of the phone opens a "Search or Ask" panel that blends traditional phone search with an AI conversation interface. Siri now lives inside the Dynamic Island, the small pill-shaped cutout at the top of newer iPhones, rather than glowing around the screen edges as before.
The Google angle is the defining detail. In January, Apple and Google announced a multi-year deal to rebuild Apple's AI on top of Google's Gemini models. Apple's own internal AI work fell behind; the head of its model team left for Meta last year. Rather than pretend otherwise, Apple licensed what it needed, the same logic it applied when it made Google the default search engine on iPhone years ago, a deal that now generates roughly $20 billion a year for Google.
The practical features are meaningful for anyone who uses an iPhone at work. Siri can now pull information from your emails and messages mid-phone-call, so if you are on with an airline it can surface your flight details without you touching the screen. It can draft messages that match how you normally write to a specific person. It understands what is on your screen and can act on it. The camera gets a new Siri mode, pointing your phone at something and asking a question, similar to what Google Lens already does.
Apple also opened up image generation tools to developers via a new programming interface, meaning third-party apps will soon be able to use Apple's photo-editing and image-creation capabilities directly. Photos edited with Apple Intelligence will carry a hidden watermark to flag AI involvement.
The geographic limits matter. Siri AI will launch in English, in beta, in the US and a handful of other countries this fall. iPhone and iPad users in the European Union will not get it at launch; local regulations are blocking it. China is also excluded. For any organization with a significant EU or Chinese workforce on iPhones, these features will arrive later than for US-based staff.
Markets were unconvinced. Apple's stock fell nearly 2% on the day, its worst single-day drop in three weeks. Investors had expected a firm September date; Apple gave them "later this year" with a beta label. One analyst called it "ticking a box, but still uninspiring." The company is also navigating a leadership transition: Tim Cook stepped down in April and John Ternus, Apple's head of hardware engineering, takes over as CEO on September 1. Ternus was not present at WWDC.
Apple settled a $250 million lawsuit in May over Siri features advertised with the iPhone 16 in 2024 that never arrived on time. When Siri AI ships this fall, still labeled beta, the credibility question is straightforward: does it work as shown, or does it become the third straight year of delayed promises?
Apple's scale remains its real advantage. ChatGPT has around 900 million weekly users. Apple has 2.5 billion active devices globally. When Apple ships a feature, it reaches people who have never opened a standalone AI app and probably never will. Whether the feature actually works as demonstrated will determine whether iOS 27 is remembered as the recovery or simply a new chapter in the same slow-motion delay.