Table of Contents
Imagine for a moment: your iPhone, instead of launching Siri by default, asked you “Which intelligence would you like to converse with today?” This scenario, still largely science fiction for many, could well become a reality by iOS 27.
According to the latest leaks from Bloomberg journalist Marc Gurman, ChatGPT’s exclusive reign on Apple’s operating systems is coming to an end. It’s no longer a question of “if,” but “when” users will be able to choose their preferred language model.
OpenAI’s AI certainly had a head start when it was first integrated. But in nearly two years, the landscape has radically changed. Players like Claude, DeepSeek, and Google Gemini have shown impressive, sometimes specialized, capabilities that deserve a place in the sun.
Exclusive ChatGPT Integration
OpenAI’s AI is the first and only to be natively connected to Siri on iOS, offering detailed responses when the assistant struggles.
Siri Powered by Google Gemini
Apple confirms the use of Google Gemini models to power a new version of Siri, signaling an initial opening.
Choice of Language Model
Users will be able to choose their preferred AI from a selection, much like a default browser or email app.
Apple’s Strategic Shift
This decision is not so surprising. The LLM market has diversified and specialized, making a monopoly unsustainable in the long run. Think of it like choosing your default search engine or browser: a basic freedom today.
Beyond mere “choice,” Apple plans to transform interaction with these AIs through “extensions.” Siri’s current use of ChatGPT remains quite basic, often serving as a simple gateway for complex questions.
Beyond Choice: The Extensions That Change Everything
With these extensions, competing language models will be able to leverage Apple Intelligence features. We’re talking about capabilities to write text, proofread it, translate it, or even generate images via the Playground app. For a developer or creative, this opens up a true Swiss Army knife of possibilities.
Multimodal Input
The user submits a request to Siri or directly via an app, blending text, images, or voice context.
Intelligent Delegation
Apple Intelligence identifies the most relevant AI based on the task and user preferences (e.g., Claude for a long text, Gemini for web integration).
Execution via Extension
The chosen AI uses its “extensions” to access system functionalities (writing, translation, image generation) and process the request.
User Feedback
The result (text, image, response) is presented to the user, potentially with a distinct voice to identify the responding AI.
Advanced users will also find dedicated modules in the Shortcuts app. Programming complex functions, automating creative or productivity tasks will become more accessible. And that, clearly, changes quite a lot for daily use.
Digital Identity at the Heart of the Stakes
One last, rather clever idea: the ability to define a different voice for each LLM used with Siri. At a time when the boundaries between artificial intelligences are blurring, knowing precisely who you are talking to becomes fundamental.
The AI Identification Challenge
With a proliferation of models and voices, the ability to distinguish the origin of an AI response is crucial for trust and bias management. Apple seems to have understood this.
This opening from Apple reflects a broader trend: the AI market is consolidating around a few giants (OpenAI, Google, Anthropic), but also seeks to diversify. Offering choice is also recognizing the value of this competition for innovation.
The iPhone, often perceived as a walled garden, is opening up. But this new freedom raises a question: does more AI choice truly mean more user control, or simply increased complexity to manage in figuring out which intelligence best serves one’s interests?
About Rigaud Mickaël
Passionate about tech and a Linux enthusiast, I decipher AI with a unique and intense vision to make it useful to all, between robots, rock and the geek universe.
