Apple Opens the Walled Garden: iOS 27 Will Let You Replace Siri with Claude or Gemini
For the first time, Apple will allow third-party AI models as the default assistant on iPhones. Users will be able to choose Gemini or Claude instead of Siri.
In the most significant opening of its platform in years, Apple has confirmed that iOS 27 will allow users to replace Siri with third-party AI models. The initial supported options: Google's Gemini and Anthropic's Claude.
I have been watching Apple long enough to know that this kind of move does not happen casually. Apple is, by institutional DNA, a company that controls the full experience. Every major opening of the iPhone, choosing a default browser, allowing third-party app stores in the EU, letting users pick a default maps app, came only after years of resistance and, usually, regulatory pressure. This Siri change is different. It is voluntary and strategic.
Apple's AI journey: a brief history
To understand why this matters, you need to understand how Apple got here. Siri launched in 2011 as the first mainstream voice assistant. It was genuinely impressive for its time. But while Google Assistant, Alexa, and eventually ChatGPT pushed the boundaries of what AI could do, Siri stagnated. The problem was not talent. Apple hired smart people. The problem was structural. Apple's privacy commitments meant Siri could not train on user data the way Google's models could. Apple's insistence on on-device processing limited model size. And Apple's culture of shipping polished, fully-baked features meant it was always slower than competitors who were willing to ship imperfect products and iterate publicly.
The "Apple Intelligence" suite announced at WWDC 2024 was supposed to fix this. It did not. The features shipped late, the writing tools felt basic compared to ChatGPT and Claude, and the much-hyped Siri 2.0 with personal context and on-screen awareness was delayed repeatedly. By early 2026, it was clear internally that Apple's AI models were not catching up fast enough.
What's changing
Currently, iPhone users can access ChatGPT, Claude, and Gemini through separate apps, but Siri remains the system-level assistant, the one that responds to button presses, voice commands, and system integrations. iOS 27 breaks that exclusivity.
Users will be able to select a default AI assistant in Settings, similar to choosing a default browser or email app. The selected assistant gets system-level access: responding to "Hey Siri" (or a new wake word), integrating with apps, and handling on-device tasks.
This is significant technically as well as strategically. System-level access means the third-party assistant can read your screen, interact with apps, send messages, set calendar events, and control device settings. Apple is exposing APIs to third-party AI providers that it has never exposed before. The security and privacy implications are enormous, and Apple says it is handling them through a combination of on-device permission prompts, sandboxed execution, and a user-auditable activity log. Whether that satisfies privacy advocates remains to be seen.
Why it matters
This is a concession from a company built on controlling the full stack. Apple's internal AI efforts, the "Apple Intelligence" suite announced in 2024, now referred to internally as Siri 2.0, have reportedly struggled to match the capabilities of frontier models from Google and Anthropic.
Rather than fall further behind, Apple is doing what it did with Google Maps and default search: acknowledging that third-party providers can do certain things better, and monetizing the integration instead. The financial terms have not been disclosed, but analysts expect Google and Anthropic are paying for default placement, similar to Google's estimated $20 billion annual payment to be the iPhone's default search engine.
For users, the benefit is clear: you get a genuinely capable AI assistant on your iPhone, not the frustratingly limited Siri experience that has been the norm for over a decade. For developers, this opens up a new integration surface. Apps will be able to register actions and intents with the user's chosen assistant, creating a more cohesive experience than the current model where each AI app is a separate silo.
Android comparison
Android users might reasonably wonder what the big deal is. Google Assistant has been replaceable on Android for years, right? Sort of. You can install alternative assistants on Android, but the integration is not smooth. System-level functions like "Hey Google" hotword detection, lock screen access, and deep OS integration have always favored Google's own assistant. What Apple is doing, giving third-party assistants true system-level parity, would actually put iOS ahead of Android in AI assistant openness. That is not a sentence I ever expected to write.
The user experience
The feature will not ship until iOS 27's public release (expected September 2026), but developer betas are already circulating. Early impressions: Claude handles writing and analysis tasks noticeably better than Siri ever did. Gemini's deep Google integration (Gmail, Calendar, Maps) makes it feel like a natural iOS assistant for users in Google's world. And Siri? It is still there for users who prefer Apple's privacy-first, on-device-only approach. But the days of Siri being the only option on an iPhone are over.
I think this is a net positive for everyone. Apple gets revenue from placement deals and keeps iPhone users from switching to Android for better AI. Google and Anthropic get access to over a billion iPhone users. And users finally get a competent AI assistant on the device they carry everywhere. The only loser here is Siri's reputation, but honestly, that ship sailed years ago.