Apple appears to have quietly prepared Siri for a future with multiple AI providers, even as it chose not to discuss that plan at WWDC, according to reporting tied to iOS 27 beta code.
The developer beta for iOS 27 reportedly includes an Extensions framework that would let Siri route requests to outside AI systems such as ChatGPT, Claude, or Gemini. The feature is said to include a settings interface and an App Store section, although both are currently disabled on Apple’s backend. Bloomberg has also reported that Apple has held talks with OpenAI, Anthropic, and Google about giving those companies access to the system.
That design would mark a shift from Apple’s earlier, more limited approach to Siri integrations. Instead of a single partnership, the company appears to be building a platform that could let several AI providers plug into the assistant, giving users more choice over which model handles a given task.
Despite the apparent groundwork, Apple left the feature out of its WWDC keynote. The presentation instead centered on Siri itself, which Apple has been rebuilding after earlier AI plans fell short.
At the event, Apple emphasized a redesigned Siri app, new personal context features, and a privacy structure that it says protects user data. The company also presented Siri AI as powered by a custom Gemini model running on Nvidia Blackwell GPUs in Google Cloud. In that context, a model-selection system for third-party AI may have complicated the message Apple wanted to send.
Apple has spent two years trying to recover from delays and setbacks in its assistant strategy. Executives have described some of the new capabilities as experimental, and internal development reportedly changed course after an earlier version did not meet the company’s standards.
Apple’s decision to hold back the announcement appears to reflect more than simple timing. One issue is Europe. Apple said during WWDC week that Siri AI will not launch in the European Union for now, pointing to unresolved talks with the European Commission over the Digital Markets Act. The company had proposed a Trusted System Agent that would allow rival assistants to access Siri AI functions without exposing sensitive device data, but EU regulators rejected that approach.
Announcing a Siri framework that welcomes third-party AI while simultaneously saying similar access is too risky for the EU would likely have created a difficult public contradiction.
Legal tensions are another factor. OpenAI is reportedly weighing possible action against Apple over its ChatGPT partnership from 2024. Bloomberg has reported that OpenAI’s lawyers are exploring options that include a breach-of-contract notice. The company reportedly expected the deal to produce substantial subscription revenue, but has objected to the way Apple integrated ChatGPT, including the need for users to explicitly say “ChatGPT” and the limited way responses appear. A broader Siri framework that places ChatGPT alongside rivals would make that dispute more pointed.
Apple’s own narrative also matters. The company is trying to convince users that its rebuilt assistant is ready after a long period of missed expectations. Adding a model picker at the same time could have weakened that pitch by making Siri look dependent on outside systems before Apple has fully established confidence in its own.
If Apple does turn the feature on, the impact could be significant. The framework would give AI companies native access to more than 1.5 billion Apple devices without requiring users to leave Siri or install separate apps. It could also make Siri a platform layer for tasks such as writing, image generation, and open-ended chat.
For now, though, the feature remains hidden in the beta. Apple has not publicly confirmed when or whether it will arrive. The code suggests the infrastructure is in place. The open question is whether Apple’s regulatory battles, legal risks, and product messaging will allow it to launch on the schedule it has in mind.