Apple is in talks with OpenAI and Anthropic to revamp Siri using third-party large language models, according to Bloomberg. This move could mark a significant shift away from Apple’s in-house AI development. If a deal goes through, Apple would embed models like Claude or ChatGPT into Siri, replacing its own struggling AI backbone.
The company has been testing custom versions of these models on its Private Cloud Compute servers to maintain user privacy and performance standards. Internally, executives are increasingly convinced that Apple’s own AI models fall short of modern expectations. Siri’s capabilities have lagged behind Android rivals, and customer frustration has grown amid delays and missed release targets.
Apple’s AI Struggles Deepen
Apple previewed advanced Siri features during its 2024 WWDC event and promised rollout with iPhone 16. But development setbacks forced the company to delay the new Siri until at least 2026. The delay triggered user outrage and multiple lawsuits, pushing Apple to rework its AI roadmap and revisit partnerships.
Despite building its own large language models, internally branded as Apple Foundation Models. The tech giant hasn’t kept pace with generative AI progress made by competitors. The current plan sees personalized Siri features launching in iOS 26, with a more powerful LLM-based Siri expected in iOS 27.
According to Bloomberg, Apple executives believe Anthropic’s Claude outperforms other options in tests and fits best with Siri’s needs. Talks with Anthropic have progressed to early financial negotiations, but the startup is demanding a multibillion-dollar annual licensing fee that increases over time. This price point has kept Apple in parallel conversations with OpenAI.
Management Turmoil and Staff Concerns
The potential shift away from internal models has triggered concern within Apple’s AI teams. Several top researchers view the move as a vote of no confidence in their work. Some have already left, while others are considering offers from competitors like Meta, which is offering up to $40 million in compensation packages.
Leadership changes have also added to the uncertainty. Siri is now overseen by Mike Rockwell and Craig Federighi after the company sidelined AI chief John Giannandrea. His influence continues to diminish, with multiple AI-related teams moved under other departments. Apple also quietly canceled its Swift Assist project for Xcode, opting instead to let developers integrate third-party AI tools like ChatGPT and Claude.
Bloomberg reports that Apple has asked both OpenAI and Anthropic to build versions of their models tailored for Apple’s private cloud, ensuring tighter privacy and performance control. While Apple plans to open on-device AI models to third-party developers, it hasn’t made the same commitment for cloud-based models. Concerns remain about infrastructure capacity and long-term reliance on external AI.
If Apple reaches a deal, it would mirror strategies used by competitors. Samsung brands its Galaxy AI features but relies on Google’s Gemini. Amazon integrates Anthropic’s Claude in Alexa+. For Apple, bringing in outside AI may be the only viable short-term path to making Siri competitive again.