Why Apple Built a Third-Party AI System for Siri but Didn't Reveal It at WWDC

The iOS 27 developer beta includes an Extensions framework, which implements a feature allowing users to switch between and use ChatGPT, Claude, and Gemini directly within Siri. Apple has already been discussing authorization for this framework with OpenAI, Anthropic, and Google.

However, Apple made no mention of this feature at this year's WWDC. Three major strategic pressures were at play here.

• Regulatory issues: In relation to the EU's Digital Markets Act (DMA), announcing a framework that allows access to third-party AI could work against Apple in negotiations with regulators.
• Legal conflicts: With OpenAI raising the possibility of a breach of contract related to the existing ChatGPT partnership and preparing legal action, revealing a feature that would weaken ChatGPT's exclusive status is a risk factor.
• Messaging strategy: Apple was focused on proving its own AI capabilities. With the Siri AI, already rebuilt on Gemini models, still in an experimental stage in terms of performance, highlighting a third-party model selection feature could undermine the brand narrative.

Currently, the Siri AI utilizes Gemini models running via Google Cloud on Nvidia Blackwell GPUs, and once the Extensions feature is activated according to Apple's future decisions, users will be able to perform tasks such as writing tools and image generation through the model of their choice.