← Portada | Mercado

Apple Watch Series 12 listens around the clock, but Secure Enclave keeps audio local

2026-09-14 · Fuente:ITWorld Korea

Apple Watch Series 12 listens around the clock, but Secure Enclave keeps audio local

A new security white paper explains how the S11-powered Live Rewind and Siri Recap convert speech to text inside an isolated hardware block, discarding audio immediately after transcription.

Apple is outfitting the Watch Series 12 and Ultra 4 with a four-feature collection of voice-driven AI it calls Audio Intelligence. The new S11 processor carries hardware capable of listening and lightly processing audio around the clock without cutting into battery life.

Two of the capabilities extend familiar functions. Always-on Shazam detects surrounding music and identifies tracks automatically, surfacing results in a Smart Stack widget; older watches offered Shazam only manually, while the new models run it in the background like Pixel phones. Sound Recognition, previously limited to HomePod and iPhone, now reaches the wrist, detecting doorbells, fire alarms and sirens with visual alerts and working standalone rather than requiring a nearby iPhone.

The other two, Live Rewind and Siri Recap, arrive in beta later this year and draw scrutiny. Live Rewind lets users double-press the Digital Crown to read the last 15 seconds of nearby speech as text, then question Siri or save it in the Siri app; Siri Recap summarizes an entire conversation in the Siri app. Whether the watch permanently records its surroundings prompted Apple to publish a security white paper on how Audio Intelligence works.

Per the document, microphone audio flows continuously into a small buffer inside the Secure Enclave, a rolling structure that overwrites older data as new input arrives — the same mechanism many Apple products use to detect a 'Hey Siri' call. The Secure Enclave is a fully segregated hardware block inaccessible to the operating system, apps and Apple itself, and also houses Face ID and Touch ID data.

A double-press of the Digital Crown sends the buffered 15 seconds over an encrypted secure transmission to the iPhone, where an AI model converts them to text inside the iPhone's Secure Enclave. The rest of iOS, apps, the user and Apple cannot access the voice data. Conversion is pure speech-to-text that neither identifies nor distinguishes speakers, and audio is discarded immediately afterward. The text displays on the watch unless the user questions Siri or saves it in the Siri app.

Etiquetas: Apple · Apple Watch Series 12 · S11 Processor · Secure Enclave · On-Device AI · Privacy


Lecturas relacionadas


萬安算交所 AIXX · AIXX · Exchange verificado · Ir al mercado