Apple has detailed the privacy safeguards built into Audio Intelligence, the microphone-driven feature launching with the Apple Watch Series 12, after acknowledging that continuous ambient listening raises surveillance concerns. The suite, powered by the new S11 chip, lets the watch transcribe recent conversations, summarise them and flag important sounds, all without needing a paired iPhone nearby for some functions. Apple says the feature matters because it pushes the boundary of what a wearable can passively monitor, while also offering a potentially valuable tool for people with hearing loss or other accessibility needs.

The Audio Intelligence suite comprises four tools: sound recognition for alerts such as sirens or crying babies; Live Rewind, which transcribes the previous 15 seconds of speech via a Digital Crown double-press; Siri Recap, an opt-in conversation summariser saved to a new Siri app; and Shazam for music identification. Apple states that no audio is ever recorded or stored, with raw sound processed and immediately deleted inside a hardware-isolated "Secure Exclave" on the S11 chip, inaccessible to apps, the user or Apple itself, and it has published an 11-page document explaining these protections. Users can opt out of each feature individually, Live Rewind sounds an audible chime and screen indicator when active, and Siri Recap avoids capturing sensitive details like financial or identification information. The Apple Watch Series 12 starts at $399 and goes on sale 18 September.