Building a competitive AI assistant proved harder than building world-class silicon.
For fifteen years, Siri has been the voice Apple placed between its users and the world — a promise of intelligence that, by most accounts, arrived late and underdelivered. Now, eleven days before its annual developer conference, Apple is rebuilding that promise from the ground up, licensing Google's Gemini AI for roughly a billion dollars a year and opening iOS to competing AI agents in a move that quietly concedes no single company has mastered the mind it once claimed to be building. The transformation arrives in the shadow of a $250 million settlement over misleading AI advertising, and the deeper question it raises is not whether the new Siri is impressive, but whether trust, once spent, can be earned back through a keynote.
- Apple's most visible product promise — a smarter Siri — collapsed so publicly that it ended in a quarter-billion-dollar lawsuit, and the company is now racing to replace what it failed to build with technology licensed from its oldest rival.
- Leaked renders show Siri being gutted and rebuilt as a full chatbot with persistent memory, document uploads, and multi-step reasoning — features competitors have offered for years — signaling that Apple is no longer leading the AI conversation, only catching up to it.
- The decision to pay Google approximately $1 billion annually for Gemini is a rare admission from a company that built its identity on vertical control: world-class chips were not enough to produce a world-class language model.
- OpenAI, already integrated into iOS, is reportedly weighing a breach-of-contract claim after Apple never meaningfully routed users toward ChatGPT — a dispute that could complicate the multi-AI ecosystem Apple is now trying to build.
- WWDC on June 8 will be the moment of reckoning: whether Apple can frame its Google dependency without embarrassment, survive questions about the ChatGPT dispute, and — most critically — make Siri perform in unscripted reality rather than rehearsed demos.
Eleven days before WWDC 2026, Bloomberg's Mark Gurman published leaked renders of a rebuilt Siri — and the transformation is stark. The assistant that has lived inside iPhones for fifteen years is being replaced by a full-featured chatbot powered by Google Gemini, complete with a dedicated app, persistent conversation history, and the ability to accept photos and documents as input. The timing is loaded: Apple settled a $250 million class-action lawsuit last year over AI capabilities it advertised but failed to deliver, and the new Siri looks fundamentally different from what was originally promised.
Three structural changes define the overhaul. Siri is getting its own standalone app — obvious in hindsight, but a genuine departure — where users can ask follow-up questions, maintain chat history, and upload files for analysis. The Dynamic Island is being repurposed as Siri's permanent home, with a swipe-down gesture revealing a new search and conversation interface that repositions the top of the screen around AI rather than notifications. And in the most surprising shift for a company built on closed ecosystems, iOS will allow users to choose between multiple AI agents — Siri, ChatGPT, Gemini, or Anthropic's Claude — depending on what Apple integrates by launch.
The decision to license Gemini for roughly $1 billion annually is the story's most revealing detail. Apple spent years building proprietary silicon to avoid exactly this kind of dependency. That it now relies on Google — a direct competitor — is a quiet admission that competitive language models proved harder to build than competitive chips. Meanwhile, OpenAI is reportedly reviewing a breach-of-contract claim against Apple, arguing that Siri never meaningfully routed users toward ChatGPT despite the iPhone's billion-plus user base. Whether that dispute reaches court remains unclear.
Apple is also adding an AI mode to the Camera app, letting users send images to third-party AI tools or reverse-image search without leaving the camera interface — a preview of the visual AI it plans to extend to camera-equipped AirPods and smart glasses in the years ahead.
Three things will matter at the June 8 keynote: how Apple explains its reliance on Google, whether journalists press on the OpenAI dispute, and whether Siri performs in unscripted live demos rather than polished renderings. Gurman has cautioned that final designs could still change, and that Apple tests multiple versions internally. The harder question is whether a company that already broke this promise once can rebuild the trust that went with it.
Eleven days before Apple's annual developer conference, Bloomberg's Mark Gurman published rendered images of what Siri is about to become—and the transformation is stark. The voice assistant that has lived inside iPhones for fifteen years is being rebuilt as a full-featured chatbot, powered by Google's Gemini AI, with a dedicated app, persistent conversation history, and the ability to accept photos and documents as input. The timing carries weight. Apple settled a $250 million class-action lawsuit last year over misleading consumers about AI capabilities it had promised but failed to deliver. The company showed off an improved Siri at WWDC 2024, advertised it heavily with the iPhone 16 launch, then quietly shelved it for more than a year. Now it's returning, but looking fundamentally different from what was originally promised.
The leaked renders reveal three major shifts. First, Siri is getting its own standalone application—something that seems obvious in hindsight but represents a genuine departure for Apple. Users will be able to ask follow-up questions without restarting the conversation, maintain chat history across sessions, and upload PDFs or images for analysis. Second, the Dynamic Island—the pill-shaped notch at the top of newer iPhones—is being repurposed as Siri's permanent home. Swiping down from the top center of the screen will reveal a new "Search or Ask" interface, and swiping further opens a chatbot-style conversation window. This is a significant change to how users navigate their phones; the center of the screen, previously reserved for notifications, now belongs to Siri. Third, and perhaps most surprising for a company that has historically controlled every aspect of the user experience, Apple is allowing multiple AI agents to coexist inside iOS. After you swipe down, you'll choose which AI handles your task—Siri, ChatGPT, Google Gemini, or Anthropic's Claude, depending on what Apple has integrated by launch.
The decision to power Siri with Google Gemini rather than Apple's own AI models is the story's most revealing detail. Apple has spent years building proprietary chips and software stacks specifically to avoid depending on outside companies. Yet the company is now licensing Gemini for approximately $1 billion annually, a public admission that building a competitive large language model proved harder than building world-class silicon. The partnership also signals that no single AI system has won the race—not even Apple's own. OpenAI, which integrated ChatGPT into iOS in 2024, is reportedly reviewing a breach-of-contract claim against Apple. The dispute centers on visibility: users had to explicitly say "ChatGPT" to trigger the feature, and Siri never automatically routed relevant queries there. OpenAI expected the iPhone's billion-plus user base to drive subscription conversions. It didn't. Whether the disagreement reaches court or settles quietly remains unclear.
Apple is also adding an AI mode to the Camera app, sitting alongside Photo, Video, and Portrait modes. Users will be able to take a picture and send it to a third-party AI or reverse-image search without leaving the camera interface. The feature replaces what Apple currently calls Visual Intelligence. The rationale extends beyond the iPhone itself. Apple is preparing users for visual AI on their phones before rolling out camera-equipped AirPods later in 2026 and smart glasses targeted for 2027. The camera changes are a preview of where the company is heading.
The $250 million settlement looms over this entire overhaul. The class-action lawsuit alleged that Apple promoted AI capabilities that did not exist at the time of purchase, saturating television and digital media to build consumer expectations. The settlement covers iPhone 15 Pro, 15 Pro Max, and the full iPhone 16 lineup purchased between June 2024 and March 2025. Eligible users can claim between $25 and $95 per device. Apple did not admit wrongdoing, but the gap between what was advertised and what shipped was real. Mark Gurman has cautioned that even a polished WWDC keynote presentation leaves an open question: can the rebuilt Siri actually undo years of damage to the Siri brand? Apple excels at on-stage moments. Making Siri consistently useful across millions of real-world use cases is a different kind of hard. It's also worth noting that iOS 27 will drop support for several older iPhone models, meaning the new Siri is built specifically for the devices that were sold on its original promise.
WWDC begins June 8 at 10 a.m. Pacific. Developer betas arrive the same day, with public release scheduled for September alongside new iPhones. Three things will matter at the keynote: how Apple frames its reliance on Google Gemini—unusual for a company to publicly depend on a direct rival—whether journalists press on the ChatGPT dispute, and whether Siri performs in unscripted live demos, not just polished renderings. Apple recently registered a new subdomain, genai.apple.com, which remains dark but signals infrastructure for whatever is coming. Final designs could still change before the keynote. Gurman himself flagged that Apple tests multiple versions internally. The question now is whether the company can deliver on promises it has already broken once.
Citações Notáveis
Can the rebuilt Siri undo years of damage to the Siri brand?— Mark Gurman, Bloomberg
Apple resolved this matter to stay focused on doing what we do best.— Apple spokesperson, on the settlement
A Conversa do Hearth Outra perspectiva sobre a história
Why does Apple need Google's Gemini at all? Doesn't Apple have the resources to build its own AI?
Apple has the resources, but apparently not the time or the breakthrough. Building a competitive large language model is different from building chips or software. Google and OpenAI have been at this longer, with more data and more iterations. Apple chose speed over control.
That seems like a huge admission for a company that's built its entire brand on privacy and keeping everything in-house.
It is. But the alternative was shipping another broken Siri and losing more credibility. The $250 million settlement already happened. They're trying to recover from that.
What about the OpenAI dispute? Is that going to blow up at WWDC?
Probably not on stage. Apple will avoid it if possible. But journalists will ask. The real issue is that OpenAI expected the iPhone's user base to drive subscriptions, and Apple buried the feature so deep that almost nobody found it. Both sides are frustrated.
So users are going to have to choose between Siri, ChatGPT, Gemini, and Claude every time they ask a question?
Not every time. But yes, you'll be able to pick. It's messy compared to what Android does—Google just made Gemini the default everywhere. Apple is trying to have it both ways: stay open to partners while keeping Siri relevant.
Can Siri actually recover from this? People have been making fun of it for years.
That's the real test. A good keynote demo doesn't mean it works when you're trying to ask it something at 7 a.m. on a Tuesday. Apple is excellent at theater. Consistency across millions of use cases is harder.