Apple Explains How New Voice AI Features Protect Privacy

Apple has detailed in a white paper how the AI-powered Siri voice features introduced at its new iPhone Duo event protect user privacy.

Siri Voice Features and Privacy Document

Apple announced new Siri AI voice features at the iPhone Duo event, including Siri Recap, Live Rewind, Sound Recognition, and Music Recognition. Alongside this announcement, the company published a document explaining how it balances user privacy with AI-based voice listening features.

The published document emphasizes that raw audio data for the new features is processed within dedicated hardware, is not saved as a file, and is inaccessible to the operating system, applications, or Apple.

Secure Exclave and S11 Chip Technology

Apple stated that the Audio Intelligence features rely on the Secure Exclave component in the S11 chip, which powers the Apple Watch Series 12 and Apple Watch Ultra 4 models. Secure Exclave acts as a hardware-isolated compartment built into silicon to process sensor data separately from the rest of the system.

Data processed and analyzed inside Secure Exclave cannot be accessed by watchOS, applications, the user, or Apple. Audio from the microphone enters this secure zone, where it is processed for speech, sound, or music before the raw audio data is converted or stored.

User Controls and Secure Transmission

Apple stated that users themselves control when and how Audio Intelligence features are activated. For example, the Live Rewind feature requires users to double-tap the Digital Crown when they want to enable it.

If data belonging to Audio Intelligence needs to be transferred to the iPhone, the process is carried out with encryption via the Secure Exclave components of both devices. Users also choose which texts to keep from Siri Recap and Live Rewind.