Key Takeaways
1. Apple is negotiating with OpenAI and Anthropic to use their large language models for Siri after struggling with its own AI development.
2. The launch of Siri’s upgrades has been delayed twice until 2026, causing frustration within Apple’s AI team.
3. Leadership changes were made in Apple’s AI team following the failure of Siri to drive iPhone upgrades and delays in the internal LLM.
4. Apple found that Anthropic’s Claude LLM performed better than its Foundation Models for enhancing Siri’s capabilities.
5. Privacy remains a top concern, with Apple exploring options to run third-party models on its own servers to maintain user privacy.
After a time of testing its own Siri AI against competitors like ChatGPT, Claude, and Google Gemini, Apple is reportedly ready to give up on its Foundation Models.
Talks with OpenAI and Anthropic
The company is currently negotiating with OpenAI and Anthropic to utilize their large language models (LLMs) for the AI features it promised for Siri since introducing Apple Intelligence nine months ago.
Sadly, Apple has struggled to implement the Siri AI upgrade using its own LLM, even on its most advanced devices like the iPhone 16 Pro Max, and has postponed it twice until 2026. The AI team seems to be either disheartened by the unclear guidance or attracted by the huge paychecks that companies like Meta and OpenAI offer to lure away their engineers.
Changes in Leadership
When the mostly unoriginal Apple Intelligence features failed to drive an upgrade cycle for the iPhone and the internal Siri LLM faced delays, Apple decided to replace the leader of its AI team and initiated a performance review. The findings suggested that the Siri AI capabilities it intended to deliver would be better served by Anthropic’s Claude LLM instead of its Foundation Models.
Next in line for performance was ChatGPT, which is why Apple is in discussions with both Anthropic and OpenAI to establish a partnership to enhance Siri with their chatbot technologies.
Privacy Concerns
Apple’s primary concern when considering a third-party service is the privacy of iPhone users. It has explored the option of running Anthropic or OpenAI’s code on its own Private Cloud Compute server clusters, asking them to create tailored ChatGPT and Claude models that would allow Apple to maintain control over the privacy settings for future Siri AI users.
Source:
Link
Leave a Reply