Following a high-profile series of failures regarding Apple’s more personalized, AI-powered Siri, the Cupertino company is reportedly thinking about giving up on using its own in-house models and partnering with the likes of Anthropic or OpenAI instead.
According to Bloomberg, Apple has already met with both companies about using their LLMs to power its smarter Siri. The companies have also been asked to train versions of their AI models that would be able to run on its Private Cloud Compute servers.
Tests have indicated that Anthropic’s Claude is the most promising AI assistant for powering Siri. As a result, Apple’s VP of Corporate Development Adrian Perica has started talking to Anthropic about what a partnership between the two companies might look like.
The report also notes that OpenAI offered to train AI models for Apple last year, but at the time, Apple was still convinced its in-house models would suffice. Now, Anthropic is reportedly asking for a multibillion-dollar yearly fee that increases every year for a similar deal. The disagreement over pricing has opened the door for OpenAI and potentially other third parties.
With that said, none of this has been finalized, and Apple hasn’t made any decisions yet about using third-party AI models to power its more personalized Siri.
Outsourcing the AI powering Siri obviously raises serious privacy concerns, which is why Apple wants to run the AI models on its own chips. By running custom versions of ChatGPT or Claude on its own servers, Apple believes it will be able to better protect user data.
This AI partnership would be a stunning admission of failure from one of the largest technology companies on the planet. Apple’s “LLM Siri” is still in active development, but was a no-show at this year’s WWDC. Meanwhile, nearly every Android vendor has charged full steam ahead with generative AI implementation — including Samsung. Apple has already fallen behind, and risks falling back even further without a reliable partner.
