Apple AI Extensions Could Make iOS 27 More Open Apple AI Extensions may let users choose Gemini, Claude, or other rival models for Siri, Writing Tools, and image features.

A logo featuring a number, inspired by the sleek design language of iOS 27.
Image Credit: AppleMagazine

Apple AI Extensions could become one of the company’s most important AI changes yet, giving iPhone, iPad, and Mac users a way to choose rival models for core Apple Intelligence features instead of relying only on Apple’s own systems or the current ChatGPT integration.

Bloomberg News reported that Apple plans to let users select third-party AI models for tasks such as generating and editing text and images across iOS 27, iPadOS 27, and macOS 27. Reuters, citing the Bloomberg report, said the feature is called “Extensions” and would allow AI providers to work through compatible App Store apps, with Apple internally testing integrations from Google and Anthropic. Apple did not immediately comment to Reuters.

The reported change would mark a sharp shift in Apple’s AI strategy. Apple Intelligence currently includes built-in features and ChatGPT access for certain requests with user permission. With iOS 27, users could reportedly choose services such as Google Gemini or Anthropic’s Claude for Apple Intelligence tasks, including Writing Tools, Image Playground, and Siri-related responses.

The timing is important. Apple is still widely seen as playing catch-up in generative AI, even as the company has emphasized privacy, on-device processing, and Private Cloud Compute as the foundation of its approach. Opening Apple Intelligence to rival models would let Apple improve capability faster without needing every model breakthrough to come from inside Cupertino.

A New Settings Layer for AI Choice

Apple AI Extensions would reportedly live inside Settings, giving users control over which outside models power specific AI tasks. That would make model choice feel more like choosing a default browser, keyboard, or search engine than downloading a separate chatbot app and remembering to open it.

According to reports, third-party providers would need to add support through their App Store apps. That is an important detail because it keeps Apple in control of distribution, privacy prompts, permissions, and system integration. Instead of letting any AI service plug directly into iOS, Apple would create a structured extension layer where users choose from supported providers installed on their devices.

The most useful version would let users assign different models to different jobs. A user might prefer one model for writing, another for image generation, and another for conversational answers. That would make Apple Intelligence more flexible without forcing Apple to pick a single outside winner.

Reports also say Apple may let users assign different voices to Siri conversations based on which model is responding. That could make the experience clearer. Siri could keep one voice for Apple’s own system actions, while Gemini, Claude, ChatGPT, or another model could use a distinct voice when handling a response.

That clarity matters because Apple needs to avoid confusion. If Siri answers sometimes as Apple, sometimes through ChatGPT, sometimes through Gemini, and sometimes through Claude, users need to know when an outside provider is involved and what data may be shared. Apple’s current ChatGPT integration asks permission before sending requests to OpenAI. A broader Extensions system would likely need the same kind of visible control.

Apple AI Extensions - Visual Intelligence in iOS 18.2 unlocking new iPhone 16 capabilities for enhanced user experience

Siri Becomes a Platform, Not Just an Assistant

The bigger implication is that Siri could become a platform for AI models rather than a single assistant trying to do everything alone. That would be a major change for Apple. Siri has long been criticized for lagging behind newer AI chatbots, but it still has one advantage none of those services can easily replicate: system-level access to iPhone, iPad, Mac, Apple Watch, and Apple’s app ecosystem.

If Apple lets rival models work through Siri, the assistant can become more useful without Apple having to build the best model for every task. The company can focus on the interface, privacy layer, device context, app actions, and user controls, while letting outside models compete on generation quality.

That strategy would also reduce risk. AI models improve quickly, and today’s best model may not remain the best for long. A model-choice system gives Apple flexibility. If Gemini becomes stronger in one category, users can choose it. If Claude performs better for writing or reasoning, users can choose that. If OpenAI improves image or voice features, ChatGPT can remain part of the mix.

For developers, the change could open a new layer of competition inside iOS. AI apps would no longer live only as standalone destinations. They could become system extensions. That means the most important AI services may compete not only for app downloads, but for default placement inside Apple Intelligence workflows.

Apple still benefits from being the gatekeeper of that experience. If the reports are accurate, users will not be replacing iOS with an outside AI platform. They will be choosing outside models inside Apple’s system. That preserves Apple’s control while making the platform feel more open.

A Strategic Answer to Google and Microsoft

Apple AI Extensions would also be a practical answer to the AI lead built by Google, Microsoft, OpenAI, and Anthropic. Microsoft has deeply integrated Copilot across Windows, Office, and enterprise workflows. Google is pushing Gemini across Android, Search, Workspace, and Pixel devices. OpenAI has turned ChatGPT into one of the most recognized consumer AI products in the world.

Apple’s advantage is different. The company controls the most valuable personal device ecosystem in the market. The iPhone is already where users communicate, search, pay, take photos, manage calendars, track health, use apps, and store personal context. If Apple can turn that device into a flexible AI hub, it does not need to win every model race by itself.

The reported Extensions approach also helps Apple avoid a one-partner dependency. ChatGPT was the first outside model integrated into Apple Intelligence, but relying too heavily on OpenAI could leave Apple exposed if user preferences, regulation, pricing, or model quality shift. Testing Gemini and Claude gives Apple more leverage and more options.

The move also fits Apple’s broader need to show AI progress under a leadership transition. Tim Cook is preparing to step aside as CEO, with John Ternus expected to inherit a company facing higher expectations around AI, Siri, hardware integration, and Services growth. Apple’s recent guidance for 14 percent to 17 percent sales growth in the third quarter gives the company financial momentum, but Wall Street is still looking for clearer AI results.

WWDC, scheduled to begin June 8, is expected to bring more details. If Apple presents Extensions as part of iOS 27, the company can frame the feature as user choice rather than AI weakness. The message would be that Apple Intelligence is becoming more capable because users can choose the model that fits the task.

Two people stand in silhouette, talking in front of a brightly lit Microsoft logo on a dark background, highlighting the growing importance of AI infrastructure in today's tech landscape.
Image Credit: REUTERS | Bruna Casas

The Risk Is Complexity

The challenge is that model choice can become confusing quickly. Most iPhone users do not want to manage AI providers the way professionals compare cloud services. They want the system to work. If Settings fills with too many model options, voices, permissions, and task assignments, the feature could feel more complicated than useful.

Apple’s job will be to make the choice simple. The default Apple Intelligence experience should work without extra setup. Advanced users can choose Gemini, Claude, ChatGPT, or another supported model when they want more control. That balance is important because Apple’s best software features usually feel powerful without demanding constant management.

Privacy will also be a central issue. Sending a Writing Tools request, image prompt, document summary, or Siri query to a third-party model is different from processing it on device or through Apple’s Private Cloud Compute. Apple will need clear permission prompts, visible provider labels, and settings that let users decide when external services can be used.

There is also a regulatory angle. Opening Apple Intelligence to rival models could help Apple answer criticism that it favors its own AI systems. At the same time, Apple will still control which apps can provide Extensions, how they appear in Settings, and what system access they receive. Regulators may watch that layer closely, especially in Europe, where platform rules already affect Apple’s software design.

The reported iOS 27 change could become Apple’s most realistic AI move: not trying to beat every rival model at once, but making the iPhone the place where those models compete. If Apple gets the interface right, Apple Intelligence becomes less about one assistant and more about a controlled AI marketplace built into the operating system.

Hannah
About the Author

Hannah is a dynamic writer based in London with a zest for all things tech and entertainment. She thrives at the intersection of cutting-edge gadgets and pop culture, weaving stories that captivate and inform.