Leak Reveals Deep Integration of Siri with ChatGPT and Claude

Screenshot 2026-09-16 142923
Leak reveals deep integration of Siri with ChatGPT and Claude
  • +
  • -
AI-powered Siri will work alongside ChatGPT or Claude for certain tasks, and third-party models may even be able to fully replace Siri.

Siri was previously integrated with certain ChatGPT models. Now, Anthropic's large language models are set to be available as AI agents on Apple devices as well.اضافة اعلان

Apple is putting major effort into making AI-powered Siri the next standout feature in its product ecosystem, and its latest event made clear that the redesigned voice assistant will be marketed as one of the key selling points of the new iPhone 18 Pro.

But the tech giant doesn't appear to intend to keep its devices confined within a closed ecosystem, away from competitors, according to a post on X from a leaker known for publishing upcoming code for iPhone and Mac operating systems.

According to the post, Apple Intelligence will feature deep integration with third-party models, including OpenAI's ChatGPT and Anthropic's Claude, with an option allowing users to fully replace Siri with certain GPT models, making them the primary agent used on the device.

Representatives for Apple, OpenAI, and Anthropic did not immediately respond to requests for comment.

The latest private frameworks reportedly leaked for iOS 27 and the "Golden Gate" Mac operating system reference the inclusion of Claude models within a model-delegation feature, which lets users of AI-powered Siri ask Apple's model to hand off certain tasks to another AI agent.

In a video the leaker posted on X, Siri passes a request to Claude to create a reminder within the Reminders app.

In this video, Claude effectively functions as the on-device AI system, while Siri acts as an intermediary. If the request requires access to Apple-specific information, AI-powered Siri handles the task itself instead.

The leak also revealed a more advanced integration for apps that fully replaces the AI-powered Siri model with a different large language model.

Using a model-management service within the private framework, the leaker was able to run GPT-5.6 using the information Siri records, and execute requests through Siri's interface and voice.

The video shows GPT-5.6 searching through emails stored on the device, identifying specific information within the inbox, then using that information to send messages to people via Apple's Messages app.

All of this information, which normally passes through AI-powered Siri, can also be logged within OpenAI's web interface, for later reference through a different platform.

So has Apple voluntarily begun dismantling its closed ecosystem? Probably not.
The European Union's Digital Markets Act, which requires Apple to give third-party developers equal access to the features of its devices and software, has played a major role in shaping how the company rolls out new operating system updates.

The legislation specifically targeted AI-powered Siri and Google's Gemini, determining that phone manufacturers would in the future need to allow third-party AI developers access to the software updates present on devices, with the aim of preventing companies from exercising excessive control over market access.

While Apple isn't opening its doors to outside AI companies out of purely voluntary motives, this move represents significant news for OpenAI and Anthropic, the developers of ChatGPT and Claude respectively.

The two companies are competing to develop more advanced models and prove their value to the public, in an environment that still lacks broad-based regulation.

With the possibility of both companies going public at massive financial valuations drawing closer in the near future, OpenAI and Anthropic are likely to continue their efforts to tap into the large market of iPhone users in the United States.

Resource: Al-Ghad