Apple Poised to Overhaul Siri With AI Upgrade at WWDC

All that data is locked down by design, the thing standing between Siri and usefulness.
Apple's privacy architecture protects user data but prevents Siri from becoming truly intelligent.

For fifteen years, Siri has lived on billions of devices without ever becoming indispensable — a quiet failure that the arrival of genuinely capable AI assistants has made impossible to overlook. At its annual developer conference, Apple now faces the central tension of its own design: it holds more intimate knowledge of its users than any competitor, yet its deepest commitment is to never use it. What unfolds this week is less a product announcement than a reckoning with the question of whether privacy and usefulness can coexist in the age of intelligent machines.

  • Hundreds of millions of iPhone users have quietly abandoned Siri for ChatGPT and Claude, making Apple's two-year-old promise of an overhaul feel like a debt coming due.
  • Apple sits on an unrivaled trove of personal data — emails, calendars, messages — that could make Siri transformative, but its own privacy architecture has kept that vault sealed.
  • The company is expected to announce a conversational Siri mode, user-controlled personal context sharing, and a developer extension framework that lets third-party apps choose their own AI model.
  • Apple is deliberately avoiding the autonomous, account-accessing AI agents gaining traction elsewhere, betting that restraint and maturity will prove more durable than speed.
  • The real verdict arrives not on Monday but in the months after, when developers and users discover whether Apple has finally found a way to make Siri know you without betraying you.

Apple arrives at its developer conference carrying a failure that has been visible for years. Siri, launched in 2011, still behaves much as it did then — and the company's 2024 promise of a meaningful overhaul produced nothing. Meanwhile, users have made their preferences clear, opening ChatGPT or Claude whenever they need something done. In markets like China, the shift has been wholesale, with AI agents handling schedules, transactions, and the small repetitive work of daily life.

The irony is that Apple holds an advantage no rival can replicate. Its 2.5 billion active devices contain the most intimate record of how people actually live — messages, calendars, notes, photos. If Siri could read that data, it would stop being a novelty and start being genuinely useful. But Apple's privacy architecture was built precisely to prevent that kind of access, and dismantling it would undermine the moral and competitive foundation the company has spent decades constructing.

Expected announcements include a conversational mode for Siri, a mechanism for users to share personal context with the assistant, and an extension framework letting developers plug their apps into Siri while choosing which AI model — OpenAI, Anthropic, or Google's Gemini — powers the experience. The structure of that framework may matter more than any single feature, as developers have been waiting to understand what Apple will actually allow them to build.

What Apple will not do is chase novelty. Tools that deploy armies of AI agents to log into accounts and execute complex tasks have drawn attention from Microsoft and Nvidia, but Apple's instinct is to wait until a technology is settled before committing to it. That caution has cost the company first-mover advantages before. This time, given unresolved security questions around autonomous agents, the patience may be well placed.

The announcement is only the beginning. Whether Apple can unlock the personal data sitting on its devices — while keeping it genuinely private — will determine whether Siri finally becomes what it always should have been.

Apple is walking into its developer conference on Monday with a problem that has been sitting in plain sight for two years: Siri, the voice assistant that shipped on the first iPhone in 2011, still feels like it did then. The company promised an overhaul in 2024 and delivered nothing. Now, with hundreds of millions of iPhone users defecting to ChatGPT, Claude, and other AI agents that actually do things, the stakes have become impossible to ignore.

The numbers tell the story. Siri lives on 2.5 billion Apple devices—nearly every iPhone, iPad, and Mac in active use. Yet people have chosen to open a different app when they need real help. In markets like China, consumers have moved wholesale to AI agents, sophisticated bots that can manage schedules, handle transactions, and automate the small repetitive work that fills a day. Apple's own customers have voted with their thumbs, and the vote was not close.

What makes this failure particularly stinging is that Apple sits atop something no other company possesses in the same way: a vault of personal data. Every iPhone contains emails, text messages, calendar entries, notes, photos, and a thousand other intimate details of how a person actually lives. That data, if Siri could access it, would transform the assistant from a novelty into something genuinely useful. A smarter Siri could answer questions by reading your own messages, manage your schedule by understanding your actual commitments, and anticipate what you need because it knows who you are.

But here is the trap Apple built for itself. All that data is locked down—not by accident, but by design. Apple's entire privacy architecture is built on the principle that apps cannot read each other's data, and that Apple itself cannot access much of it without explicit permission. This is the company's competitive moat and its moral claim. It is also the thing standing between Siri and usefulness.

Analysts expect Apple to announce a "chat" mode for Siri on Monday, along with a way for users to share personal context with the assistant. The company will likely open Siri to third-party developers through what it calls extensions, letting them plug their own apps into the assistant and choose which AI model powers them—OpenAI, Anthropic, or Google's Gemini. Apple may also introduce new ways for developers to tap into the processing power of its custom chips. The framework matters as much as the feature. Developers have been waiting to see how Apple will let them build AI experiences without dismantling the privacy walls that define the platform.

What Apple will not do, at least not yet, is chase the cutting edge for its own sake. The company has never worked that way. It will frame AI not as a technology but as a set of experiences—things that make your phone more helpful, more attentive, more yours. This matters because Americans remain uneasy about AI, and Apple knows it. The company's customers in other markets are more optimistic, but Apple has built its brand on the principle that it does not adopt technology just because the technology exists.

There are emerging tools like OpenClaw, which can direct armies of AI agents to log into your accounts and carry out complex tasks. Microsoft and Nvidia have spent time on it. Apple is unlikely to follow. The security questions are still unsettled, and Apple's instinct is to wait until a technology is mature before betting the company on it. That caution has sometimes cost Apple the first-mover advantage. This time, it may be exactly right.

The real test comes after Monday. Apple can announce features. What matters is whether unlocking that personal data—while keeping it actually private—will finally make Siri the thing it should have been all along.

They have to make Siri not suck, but Apple also has to put the framework together of how their developers can take advantage of AI themselves. AI is all about data, because data is what creates context and what creates better results.
— Patrick Moorhead, founder of Moor Insights & Strategy
Apple is likely to frame AI not as a technology but rather as experiences or features that its customers will find helpful.
— Andrew Cornwall, senior analyst at Forrester
A Conversa do Hearth Outra perspectiva sobre a história
Inventor

Why does it matter that Siri hasn't changed since 2011? Isn't it still functional?

Model

Functional and useful are different things. Siri can set a timer. But ChatGPT can read your emails and help you draft a response. That's the gap.

Inventor

So Apple's problem is that it promised to fix Siri and didn't?

Model

That's part of it. But the deeper problem is that Apple has all this personal data on your phone—your calendar, your messages, everything—and Siri can't use it. A competitor's AI can, because it lives in the cloud.

Inventor

Why doesn't Apple just let Siri access that data?

Model

Because Apple built its entire business on the idea that your data stays on your device. If Siri reads your emails, Apple has to make sure no one else can. It's a privacy feature that became a competitive disadvantage.

Inventor

So what's Apple announcing on Monday?

Model

Probably a way for you to opt in—to tell Siri it's okay to read your personal stuff. And a way for other apps to plug into Siri, so developers can build AI features without Apple doing all the work.

Inventor

Will that be enough to catch up?

Model

It depends on execution. The technology is there. The question is whether people trust Apple enough to hand over that data, and whether developers actually build on top of it. If they do, Siri could finally become something worth using.

Quer a matéria completa? Leia o original em Rappler ↗
Fale Conosco FAQ