Overview
- Apple introduced a rebuilt Siri AI and a second‑generation Apple Foundation Model at WWDC as part of its Apple Intelligence effort.
- The system uses on‑device models for routine requests and a Private Cloud Compute layer for more complex tasks, a design Apple says preserves user privacy.
- Reporting shows some cloud inference will run on Nvidia GPUs inside Google Cloud data centers under confidential compute, a reliance Apple has not fully detailed.
- Apple framed Siri AI as a productivity tool that will refuse romantic or companion requests, stressing utility over engagement.
- Early reviewer testing praised practical features but also reported discomfort with personal‑context access, and several advanced functions will only run on the newest high‑end Apple devices.