By Chinenye Anuforo
Apple is reportedly in discussions with Anthropic PBC and OpenAI to integrate their advanced large language models (LLMs) into a future version of Siri. The potential collaboration suggests a significant re-evaluation of Apple’s in-house artificial intelligence strategy, aiming to bolster Siri’s capabilities and bring them on par with other leading AI assistants.
The iPhone maker is exploring the possibility of running these third-party models on its own Private Cloud Compute servers, utilizing its powerful Mac chips to ensure user privacy and data security. While Apple has historically relied on its own “Apple Foundation Models” for most of its AI features, a move to external LLMs like Anthropic’s Claude or OpenAI’s ChatGPT for Siri would mark a substantial deviation from its previous plans for a 2026 update, which were set to be powered entirely by its homegrown technology.
This strategic consideration comes as Siri, first introduced in 2011, has fallen behind the curve compared to more sophisticated AI chatbots. Past efforts to enhance the voice assistant have faced engineering hurdles and delays, pushing planned improvements to next spring. Internally, Mike Rockwell, who recently took over Siri engineering, and software engineering head Craig Federighi initiated an assessment to determine whether Apple’s in-house AI or third-party solutions would best serve Siri’s evolving needs. Testing indicated that Anthropic’s technology was particularly promising.
The potential shift has created some apprehension within Apple’s AI teams. Some engineers are reportedly concerned that considering external technology undermines their efforts and creates a perception of responsibility for the company’s AI shortcomings. This has led to worries about talent retention, especially given the significantly higher compensation packages offered by competitors like Meta.
Financial negotiations are also a key factor; Anthropic is reportedly seeking a multi-billion dollar annual fee that would increase over time. This could prompt Apple to consider alternatives like OpenAI if an agreement isn’t reached.
Should Apple proceed with a third-party partnership, it could further reduce the influence of John Giannandrea, Apple’s AI chief and a known advocate for in-house LLM development. His responsibilities have already been curtailed, with several AI-related teams now reporting to Federighi. Even internal projects, such as “Swift Assist” designed to help developers with coding, have been discontinued in favor of integrating third-party programming models like ChatGPT or Claude.
Despite ongoing internal uncertainties, Apple’s leadership appears increasingly open to leveraging external technology for a rapid improvement in its AI offerings. While the company continues to develop its own AI models for on-device and developer use cases, the exploration of third-party integration into Siri signals a significant strategic re-evaluation of how Apple intends to compete in the fast-paced generative AI landscape.
Leave a comment