Apple is Reportedly Working with Google on a Siri AI Model

Last year, Apple released its first Apple Intelligence features using OpenAI ChatGPT as a fallback of sorts. At the time, Apple said that it was looking into partnering with other AI companies as needed, and Google specifically came up as a possibility. But a new report in Bloomberg suggests that the Apple/Google AI partnership could be even further reaching.

Mark Gurman claims that Apple is exploring whether it makes sense for Google to create a custom Gemini-based AI model for its Siri assistant to finally realize the conversation Siri capabilities it first announced at WWDC 2024 but never delivered. Google has started training a model that would run on Apple’s Private Cloud Compute servers, though the talks between the two companies are still preliminary.

Gurman says that Apple has also explored partnerships with Anthropic and OpenAI to potentially use their AI models for Siri. Anthropic was the front-runner with Craig Federighi and other executives, but the financial terms it demanded were too high for Apple so it expanded the search.

Google is separately working with Apple to integrate Gemini more broadly into Apple Intelligence in a manner similar to today’s ChatGPT hand-off. And Apple abandoned plans to use an in-house model for its Xcode developer tool: It will now use models from OpenAI and Anthropic there instead.

Apple has two AI-based Siri efforts underway for now, one codenamed Linwood that would be powered by its own models and the other, Glenwood, that would work with external models. It will proceed with one or the other depending on which better meets its needs.