A feature that turns the last few seconds of nearby speech into on-screen text has landed on millions of wrists, and not everyone within earshot agreed to it.
The Apple Watch Series 12 and Watch Ultra 4 shipped with Audio Intelligence, a set of four sound-processing tools running on the S11 chip. Sound Recognition flags sirens, alarms and doorbells. Music Recognition leans on Shazam. Live Rewind and Siri Recap are the ones drawing scrutiny.
Live Rewind fires only when the wearer double-presses the Digital Crown. The watch streams audio from its microphone, processes it inside a Secure Exclave, then hands a 15-second segment to a paired iPhone for speech-to-text. The resulting text is kept, the audio is discarded, and a tone sounds even in silent mode alongside an on-screen animation meant to warn bystanders.
Siri Recap behaves differently. Apple says it creates no recording, produces no verbatim transcript and never attributes speakers. It returns a short summary of the day’s conversations, with no audible cue because no raw audio survives and nothing can be reconstructed from it.
Both are opt-in and switchable from the iPhone Watch app or Control Center, and Recap can be scheduled by time and place.
Consent law trails the hardware
Eleven US states require every party in a private conversation to agree before it is recorded. Activists say a chime most people cannot identify is not agreement. The Electronic Frontier Foundation’s Adam Schwartz argued that conversational privacy has to include freedom from other people documenting what we say without clear opt-in.
Apple frames the design as respect for those nearby, citing the tone and the ephemeral text. Critics counter that normalizing an always-listening wrist sets a precedent that is hard to walk back.
