With the release of the Apple Watch Series 12, Apple has introduced Audio Intelligence, a suite of AI-powered audio processing features tuned for the S11 chip. Two features in particular have drawn scrutiny: Live Rewind and Siri Recap.

Live Rewind lets the wearer double-press the Digital Crown to capture the last 15 seconds of a conversation as text. The audio is processed in a Secure Exclave on the chip, then sent to the user's nearby iPhone for speech-to-text conversion. Apple says the audio is discarded after transcription. The Watch emits an audible chime even in silent mode and shows an on-screen animation to alert those nearby.

Siri Recap goes further: it "summarizes conversations throughout your day" and produces Apple Intelligence-generated notes. Apple's technical documentation states that no raw audio is retained, no verbatim transcript is created, and no speakers are identified. Unlike Live Rewind, there is no audible signal because Apple says no raw audio is stored.

Apple frames these features as respectful of privacy. In a section titled "How Live Rewind respects those around you," the company points to the chime and visual cue. For Siri Recap, it emphasises that the output is comparable to notes a person might jot down after a conversation and cannot be reconstructed into original audio.

But the Electronic Frontier Foundation (EFF) is not persuaded. "Our right to conversational privacy must include freedom from other people, without our clear opt-in consent, using technology to document what we are saying," said EFF privacy litigation director Adam Schwartz. "Otherwise, people will self-censor, and conversation will lose its spontaneity and intimacy."

Schwartz noted that bystanders have no practical way to consent to or decline the recording. He suggested people "think twice before using this technology, out of respect to the conversational privacy of others." The EFF also pointed out that it remains to be seen how state eavesdropping laws will apply to the new features.

The issue is most acute in the 11 US states that require all parties to consent to a recording. Even with the audible tone, bystanders may not recognise its meaning — and they have not explicitly agreed to be captured. Apple's approach, while less intrusive than some rival products such as AI glasses, still represents a significant expansion of always-on audio monitoring in consumer devices.