At the iPhone Duo announcement event, Apple unveiled "Audio Intelligence," a new suite of voice AI features for the Apple Watch. These features include Sound Recognition for detecting ambient sounds, Music Recognition with Shazam, Live Rewind for reviewing the last 15 seconds of speech in text, and Siri Recap for creating conversation summaries. These will be supported on Apple Watch Series 12 and Apple Watch Ultra 4, with beta versions expected to be available in English later this year.
Product LaunchesAppleAudio IntelligenceApple Watch
Apple Announces Audio Intelligence: Privacy-Focused Voice AI Features for Apple Watch
This article is a translation. Read the Japanese original
According to Apple's public documentation, these features are designed for privacy protection, and voice data is neither recorded nor stored. Voice is processed in a hardware region isolated from the system called the "Secure Exclave," integrated into the Apple Watch's S11 chip. The system is designed so that data processed in this region cannot be accessed by watchOS, apps, or even Apple itself.
Privacy considerations are also evident in data synchronization and processing. When text data from Siri Recap or Live Rewind is synced across devices via iCloud, end-to-end encryption is applied. Apple explained that even when Private Cloud Compute (PCC) is used to generate Siri Recap summaries, data is used only for temporary processing and is not stored by or accessible to Apple.
Sources
- Read the Apple document explaining how new listening features still protect your privacy （The Verge AI, 2026-09-10）
- Apple Official Support