Apple's Siri AI Can Be Swapped Out for Claude, ChatGPT, Code Shows
summary
Leaked code in iOS 27 and macOS Golden Gate indicates that Apple has built a deep infrastructure for Siri to integrate with third-party AI models like Claude and ChatGPT. This architecture allows external models to either act as extensions or completely replace Apple's server-side AI processing. Such a move likely stems from a need to comply with the European Union's Digital Markets Act, which demands fairer access to system features for third-party services. By doing this, Apple enables a more flexible AI ecosystem where users can choose their preferred LLM for system-level tasks.
key points
- The Model Delegation mechanism allows third-party AI, such as Claude, to function as a Siri extension that can interpret natural language and hand tasks back to Siri for system execution.
- A more advanced Inference Providing protocol allows an external model, like GPT-5.6, to replace Apple's server-side Siri model entirely, gaining access to system tools and personal data to perform complex actions.
- The European Union's Digital Markets Act is a primary driver for these changes, as it requires Apple to provide third-party developers with the same hardware and software access available to its own services.
who should read it
Tech enthusiasts and Apple users interested in how AI integration is evolving on iOS and macOS. It also appeals to developers and policy analysts tracking Apple's compliance with EU regulations.
discussion skeptical
consensus
Commenters generally agree that voice assistants like Siri and Alexa have failed to meet their potential due to poor UX, rigid command structures, and unreliable execution of basic tasks. While some find value in simple utility like timers or hands-free queries, the majority view the 'talking computer' trope as an inefficient interface for complex work. There is a shared sentiment that these tools have been hampered by a lack of discoverability and a failure to iterate effectively on core functionality.
top insight
The failure of voice ecosystems like Alexa Skills was not due to a lack of user desire for automation, but rather the high friction of setup and the 'death by a thousand cuts' caused by unstable third-party integrations. A significant barrier is that users are unwilling to perform the manual labor of installing specific hardware and apps just to achieve a marginally more convenient interaction. This suggests that for AI agents to succeed, they must be integrated at the infrastructure level (e.g., in new home construction) to eliminate setup friction.
Generated by an LLM from the article and thread. Check the sources before relying on it.