Apple’s new AI features for Watch Series 12 raise eavesdropping concerns

Apple just gave its wristwatch ears. And not everyone is comfortable with that.

The company’s new Apple Watch Series 12, announced September 9 at its “Surprise and Shine” event, ships with a suite of features Apple calls Audio Intelligence. The headline tools: Siri Recap, which passively listens to conversations and generates AI-powered summaries, and Live Rewind, which lets users double-press the Digital Crown to pull up a transcript of the last 15 seconds of audio. The watch starts at $399 and begins shipping September 18.

Apple insists the features are private by design. Legal experts aren’t so sure.

How Audio Intelligence actually works

Both features run on the new S11 chip, which includes a component Apple calls the Secure Exclave. Raw audio is overwritten on the spot, never saved as a recording. What remains are brief text summaries or transcripts, stripped of speaker attribution.

Siri Recap can be toggled on manually or scheduled for specific times, like a morning standup or a doctor’s appointment. It filters out content Apple deems sensitive before generating its summary. Live Rewind, meanwhile, works more like a dashcam for sound: it captures a rolling 15-second buffer that only becomes accessible when the user explicitly triggers it.

Apple published an 11-page privacy overview detailing exactly how these safeguards operate. The company wants to make very clear that no audio leaves the device, no recordings are stored, and no data hits Apple’s servers.

Live Rewind also emits an audible chime and displays visual indicators when activated, even if the watch is in silent mode.

The legal minefield

Roughly a dozen US states operate under “all-party consent” statutes for recording conversations. In states like California, Illinois, and Massachusetts, every person in a conversation generally must agree to being recorded for the recording to be legal. The fact that a watch beeps doesn’t constitute informed consent from the other people in the room.

Legal experts have flagged this as a genuine liability risk, not just for Apple, but for the millions of users who might activate these features without fully understanding the legal landscape in their jurisdiction.

Apple’s counter-argument rests on a technical distinction: these features don’t produce “recordings” in the traditional sense. There’s no audio file. There’s no playback. Just a text summary. Whether courts will find that distinction persuasive is an open question.

Why the EU gets to wait

Notably, Audio Intelligence features won’t be available in the European Union at launch. The initial rollout is limited to English speakers outside EU markets, with updates expected in future releases.

That’s almost certainly a nod to the EU’s stricter regulatory environment. The General Data Protection Regulation imposes far more rigorous requirements around consent and data processing than US federal law does. This strategy isn’t new for Apple. The company delayed its Apple Intelligence rollout in the EU previously, citing regulatory uncertainty around the Digital Markets Act.

What this means for the wearable market

The features are set to launch in beta later in 2026, giving Apple some runway to observe how users and regulators respond before a full production release.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.