In its just-concluded presentation, Apple announced that Apple Watch Series 12 and Ultra 4 are introducing all-new audio intelligence features, including Sound Recognition, Music Recognition, Real-Time Playback, and the new Siri Recap. Once enabled, the devices will be able to continuously sense and analyze the sound environment around the user.
Apple explained how Siri Recap handles conversations after listening to them.
Among them, Siri Recap is one of the most talked-about features in this update. It can continuously understand conversations around the user in everyday life and automatically organize key summaries through Apple Intelligence technology. Apple says these summaries are designed to help users “jog their memory,” allowing people to focus more on the present without worrying about missing important information.
Because this feature involves analyzing ambient audio over extended periods, Apple also anticipates that people may have privacy concerns, so it has specifically disclosed the relevant privacy protection mechanisms and how they work:
- No speaker identification
Siri Recap does not identify or label the speakers in a conversation. Although names may appear in summaries, it does not directly link content to a specific individual. - Automatically exclude sensitive information
The system filters sensitive content that may involve self-harm, hate speech, financial information, government-issued identification data, authentication credentials, and certain personal identity information. - The summary content may be incomplete.
Apple notes that AI-generated summaries may be incomplete or inaccurate, and filtering mechanisms cannot guarantee the complete removal of all sensitive information. - No full verbatim transcript is provided.
Siri Recap doesn’t create a complete conversation record; instead, it generates a content summary similar to meeting notes or key takeaways. - Keep only the core information.
The system first condenses the conversation content, then transmits it in encrypted form to a private cloud computing platform. The condensed content is usually less than half of the original information, and it removes modal particles, redundancies, and unnecessary details, retaining only the core meaning. - Does not record or store audio
All audio analysis is performed within a secure isolated environment on Apple Watch and iPhone. No audio is recorded, and no raw audio data is stored. Neither Apple, third-party apps, nor users themselves can directly access the contents of that isolated environment.
- Adopt private cloud computing.
The summarization process is powered by Apple Foundation Model and Private Cloud Compute architecture. Apple emphasizes that neither the company itself nor any third party can access the relevant data. - Adopts an active opening mechanism.
Siri Recap is off by default, and users must manually enable it in the Apple Watch app. It can also be scheduled for specific periods—such as on during work hours and off at night—and Control Center provides a quick toggle. - Summary automatically deleted
If the summary is not saved to the Siri App, the system will automatically delete it after seven days. - Supports cross-device management
Saved summaries sync to the Siri app and can be deleted or managed on all synced devices, and can also be exported to other apps such as Notes. - Full end-to-end encryption
Summary data synced via iCloud is end-to-end encrypted, and only devices authorized by the user can access it.
Apple also reminds users that when enabling Siri Recap, they should still respect the privacy of those around them. The official documentation states that even though the system does not record or save audio, and does not display the speaker’s identity, users still need to be mindful of their surroundings and potentially sensitive topics.
Siri Recap requires a new-generation Apple Watch and iPhone 16 or later models (excluding iPhone 16e), and is expected to officially launch later this year.
Source: KOCPC Chinese